id
stringlengths 20
20
| year
int64 2.01k
2.02k
| title
stringlengths 21
249
| sections
stringlengths 4.88k
32.5k
| headings
stringclasses 217
values | abstract
stringlengths 32
4.12k
| summary
stringlengths 227
2.45k
| keywords
stringlengths 7
1.31k
⌀ | toc
stringclasses 267
values |
|---|---|---|---|---|---|---|---|---|
journal.pgen.1006687 | 2,017 | Reduced dosage of β-catenin provides significant rescue of cardiac outflow tract anomalies in a Tbx1 conditional null mouse model of 22q11.2 deletion syndrome | The 22q11 . 2 deletion syndrome ( 22q11 . 2DS ) , also known as velo-cardio-facial syndrome ( MIM# 192430 ) or DiGeorge syndrome ( MIM# 188400 ) is a congenital malformation disorder that is caused by a hemizygous 1 . 5–3 million base pair ( Mb ) deletion of chromosome 22q11 . 2 ., It occurs with a frequency of 1:1 , 000 fetuses 1 and 1:4 , 000 live births 2 ., Approximately 60–70% of affected 22q11 . 2DS individuals have congenital heart disease ( CHD ) due to malformations of the aortic arch and/or cardiac outflow tract 3 ., There are over 46 known coding genes in the 3 Mb region , including TBX1 ( T-box 1; MIM# 602054 ) , encoding a T-box containing transcription factor 4 ., TBX1 has been considered the strongest candidate gene for CHD , based upon studies of mouse models 5–7 and discovery of mutations in some non-deleted patients 8 , 9 ., The basis of variable phenotypic expression is under intense investigation ., Understanding responsible genetic factors upstream and downstream of TBX1 is necessary to test for relevancy as modifiers in human 22q11 . 2DS patients ., We are taking mouse genetic approaches to identify genes and networks that may act as modifiers ., Tbx1 heterozygous mice have mild aortic arch anomalies or ventricular septal defects , at reduced penetrance , while all homozygous null mutant mice die at birth and have a persistent truncus arteriosus ( PTA ) , which is the most serious heart defect that occurs in 22q11 . 2DS patients 5–7 ., In mammals , Tbx1 is expressed strongly in the embryonic pharyngeal apparatus , but not the heart tube itself suggesting that its critical functions are in this tissue 4 ., In the early vertebrate embryo , the heart forms as a bilateral cardiac crescent of mesodermal cells , termed the first heart field that fuses to form the primitive heart tube 10 , 11 ., Additional mesodermal cells derived from the pharyngeal apparatus , referred to as the second heart field ( SHF ) migrates and helps to expand the heart tube in both directions 12 13 13–16 ., These cells remain in a progenitor state , allowing them to migrate and build the length of the heart tube , where they differentiate into smooth and cardiac muscle and endothelial cells 17 , 18 ., The SHF itself , can be further subdivided to the anterior heart field ( AHF or anterior SHF ) forming the cardiac OFT and right ventricle as well as the posterior SHF forming the inflow tract , respectively , based upon gene expression and cell lineage studies 19–21 ., Of interest , Tbx1 is strongly expressed in the pharyngeal mesoderm , including the AHF , but it is not noticeably expressed in the posterior SHF or heart tube 22–24 ., One of the key functions of AHF cells is to maintain a progenitor cell state and to prevent premature differentiation ., 25 Gene expression profiling of the AHF , within pharyngeal arches two to six , in Tbx1-/- embryos versus wild type littermates 24 and embryonic stem cell lineage studies 22 , suggest that Tbx1 serves to restrict premature differentiation of the pharyngeal mesoderm , so as to allow the OFT to elongate properly 25 ., However , the tissue specificity and key molecular mechanisms are not well defined ., The basis for premature differentiation in the AHF in Tbx1 mutant embryos is unknown ., Major signaling pathways likely have a role in this process ., The canonical Wnt signaling pathway is mediated by β-catenin , which has critical functions in most aspects of embryonic development ., There are multiphasic functions of Wnt/β-catenin in the pharyngeal mesoderm required for heart development 26 ., Several years ago , it was shown that canonical Wnt/β-catenin has a major role in the AHF in forming the cardiac OFT 27 ., Further , one study showed that increased or decreased Wnt/β-catenin in the pharyngeal mesenchyme ( DermoCre ) resulted in a decrease or increase in Tbx1 expression , implicating antagonistic functions upstream of Tbx1 28 ., However , genetic interaction studies were not explored nor were gene expression profiling performed to understand possible molecular connections ., Such studies would provide possible modifier genes to investigate in human 22q11 . 2DS to understand its variable expressivity ., In this report we performed genetic rescue experiments between Tbx1 and β-catenin in the AHF , using mouse models ., Wnt/β-catenin and Tbx1 are expressed in the opposite domains of the SHF , with Tbx1 higher in the AHF and Wnt/β-catenin higher in the posterior SHF , as denoted by Wnt2 and Mef2c-AHF-Cre 18 lineage compared to canonical Wnt signaling ( Fig 1A–1E ) ., We were interested in further exploring the function of β-catenin when completely diminished ( Mef2c-AHF-Cre/+;β-cateninflox/flox , referred to as β-cat LOF 29 ) or constitutively active ( Mef2c-AHF-Cre/+;β-cateninEx3/+ 30 , referred to as β-cat GOF ) in the AHF ., To identify downstream genes affected by these changes , gene expression profiling was performed on the distal pharyngeal apparatus containing the AHF micro-dissected from β-cat LOF and β-cat GOF embryos at E9 . 5 ( Fig 1F–1H ) ., Note that the dissection of the AHF did not include the heart tube ., In order to highlight the genes with the greatest fold change , we created a dot plot of log2 fold changes ( Fig 1I ) ., Loss of both β-catenin alleles in the Mef2c-AHF-Cre domain resulted in strongly reduced expression of muscle structural genes in the AHF , while constitutive activation of β-catenin in this domain , had the opposite effect and caused a strong increase in expression of the same genes in the AHF ( Fig 1I ) ., This increase was strikingly similar to that in the AHF of global Tbx1 null mutant embryos that were previously reported 22 , 24 , 31 ., We then examined cardiac phenotypes upon inactivation of Tbx1 in the Mef2c-AHF-Cre lineage to determine whether Tbx1 had a specific role in the AHF ., To determine a specific role of Tbx1 in the Mef2c-AHF-Cre domain , we generated two different genotypes , Mef2c-AHF-Cre/+;Tbx1f/- and Mef2c-AHF-Cre/+;Tbx1f/f ., Embryos at E14 . 5 with both genotypes had a persistent truncus arteriosus ( PTA ) with complete penetrance ( n = 50; Fig 2A ) ., Most but not all had an accompanying ventricular septal defect ( VSD; n = 30; Fig 2B and 2C ) , in contrast to Tbx1-/- mutant embryos , which all has a PTA with a VSD ., The PTA was observed as early as E12 . 5 ( S1 Fig ) ., Due to the similarity in phenotype ( Fig 2A ) , the two genotypes were combined and further referred to as Tbx1 LOF ., In Tbx1 LOF embryos at E9 . 5 , the pharyngeal apparatus and individual arches within appeared grossly normal ( Fig 2F–2G ) compared to control littermates ( Fig 2D and 2E ) ., This is distinctly different as compared to the global Tbx1-/- null mutant embryos or mesoderm specific Tbx1 conditional loss of function embryos at this stage 32 that have a severely hypoplastic distal pharyngeal apparatus ., This rules out extreme morphology defects , such as absence of neural crest cell populations , as being responsible for the presence of a PTA in Tbx1 LOF embryos ., We also performed lineage tracing ( Fig 2D–2G ) and observed that the Mef2c-AHF-Cre lineage in the AHF was only slightly reduced in Tbx1 LOF embryos versus control littermates at E9 . 5 ( Fig 2H ) ., By in situ hybridization analysis , Tbx1 expression was greatly reduced in Tbx1 LOF embryos ( S1 Fig ) and this was confirmed by qRT-PCR ( Fig 2I ) ., Cell proliferation and apoptosis in the Tbx1 LOF versus control embryos did not show any significant difference in the Mef2c-AHF-Cre lineage in the AHF region between groups at E9 . 5 ( S2 and S3 Figs ) ., This is different than what was previously found for Tbx1-/- 33 or Nkx2-5Cre 22 conditional mutant embryos , which have changes in proliferation and apoptosis ., We suggest the improved appearance of the distal pharyngeal apparatus in Tbx1 LOF embryos is due to differences in the Mef2c-AHF-Cre recombination domain ., In relation to β-catenin , we noted a slight decrease in Tbx1 expression in β-cat GOF embryos ( S4 Fig ) ., This was consistent , although not as dramatic , as was found previously using a broader mesenchymal Cre driver ( DermoCre ) 28 ., We found β-catenin mRNA is significantly increased in expression in the AHF of Tbx1 LOF embryos by qRT-PCR ( Fig 2I ) ., As for β-catenin gain of function in the AHF , we were interested in determining the function of loss and gain of Tbx1 in the AHF ., We previously generated a tissue specific constitutively expressing Tbx1 gene 34 ., Homozygous mice were crossed with Mef2c-AHF-Cre mice to overexpress Tbx1 in the same domain as other alleles , and the embryos are referred to as Tbx1 GOF ., Gene expression profiling of Tbx1 LOF and GOF embryos was performed of AHF tissue at E9 . 5 , to test whether loss or gain of Tbx1 would have opposing effects on muscle structural protein differentiation genes and to compare with findings of β-catenin loss and gain mutant embryos ( Fig 3A–3C ) ., The dot plots of global gene expression changes in the AHF between β-cat GOF versus Tbx1 LOF embryos showed increase in gene expression in the same direction ( Fig 3A ) ., The genes with the largest increase were the muscle structural protein genes ., Similarly , β-cat LOF versus Tbx1 GOF showed the same strong decrease of expression of muscle differentiation genes ., The genes with the largest decrease were the muscle structural protein genes ., These results provide functional genetic insight as to the previously implicated antagonistic relationship between Tbx1 and β-catenin in the pharyngeal mesenchyme 28 , that they perhaps are needed to balance cell differentiation ., Some of the genes were tested by qRT-PCR for Tbx1 LOF and β-cat LOF embryos ( Fig 3C ) ., We found top genes decreased in expression in Tbx1 LOF embryos were not generally decreased in β-cat LOF embryos as top genes that were increased in expression ., Tbx1 was slightly increased in expression in β-cat LOF embryos ( Fig 3C ) ., Based upon the opposing gene expression changes between Tbx1 and β-catenin in the AHF , that also included β-catenin mRNA itself in Tbx1 LOF embryos ( Fig 2I ) , we tested whether we could rescue heart defects in the Tbx1 conditional loss of function mutant embryos by inactivating one allele of β-catenin in the Mef2c-AHF-Cre domain ( Mef2c-AHF-Cre/+;Tbx1f/- and Mef2c-AHF-Cre/+;Tbx1f/f ) ., Details of the background and crosses are provided in the Methods section and details of the control genotypes are provided in S1 Table ., Inactivation of one allele of β-catenin in the Mef2c-AHF-Cre domain did not result in any cardiovascular defects ( S1 Table and 27 ) ., Significant rescue ( p < 0 . 001 , Fisher’s exact test ) was obtained in both sets of double mutant rescue genotype embryos ( Fig 4 ) ., Upon combining all double mutant embryos together ( n = 56 ) , a total of 61% ( n = 34/56 ) showed some rescue of the PTA phenotype ( Fig 4 ) ., Specifically , complete distal OFT and partial proximal OFT septation and/or complete septation between the ventricles were present in these hearts ( Fig 5A–5D ) ., Ten percent showed complete rescue ., Additional and more posterior sections can be found in S5 Fig Lineage tracing of the double mutant rescue genotyped embryos showed no significant difference in the number of cells in the AHF , of Mef2c-AHF-Cre lineage compared to the control or Tbx1 LOF embryos at E9 . 5 ( Fig 5E ) ., Finally , qRT-PCR was performed and Tbx1 and β-catenin mRNA expression were reduced in the AHF of these embryos compared to the control ( Fig 5F ) ., Since we identified the greatest increase of expression in Tbx1 LOF and decrease in β-cat LOF embryos pertaining to muscle differentiation genes , we tested if there is global normalization of expression in the embryos of the double mutant , rescue genotype ., For this test , gene expression profiling was performed on these embryos , in the same way for the individual mutant embryos and we found this to be the case ., Expression of genes with greatest increase in Tbx1 LOF embryos ( >1 . 3 fold ) , primarily the differentiation genes and greatest decrease in β-cat LOF embryos were largely normalized in rescued embryos ( Fig 6A ) ., However , we did not observe this same strong finding for genes increased in β-cat LOF embryos ., As mentioned , most of the genes with the strongest increase of expression in Tbx1 LOF and decrease in β-cat LOF embryos that were normalized ( p<0 . 01 ) in rescued embryos , were genes that encode smooth or cardiac muscle genes ( Figs 3C and 6B ) ., This also included major transcription factors such as Pitx2 , Tbx5 , Gata4 and Gata6 , that are required for cardiac muscle differentiation 35–37 ., The canonical Wnt gene , Wnt2 , showed a similar pattern ( Figs 3C and 6B ) ., A full heatmap of the experiment is shown in S6 Fig Some additional genes of note , increased in expression in Tbx1 LOF embryos by gene profiling include Myocd , Bmp2 , Bmp10 , Erbb4 and Sfrp5 , which were oppositely affected in β-cat LOF embryos , and normalized in rescued embryos ( Fig 6B and S6 Fig ) ., This data supports the idea that pro-differentiation by canonical Wnt/β-catenin might be modulated by Tbx1 in the Mef2c-AHF-Cre lineage ., Not all genes with increase in expression in Tbx1 LOF embryos and decrease in β-cat LOF showed normalization in rescued embryos ( Hand1 , Zfpm2 , Smarcd3 and Tbx20; Fig 6B and S7 Fig ) ., Further , genes reduced in expression in both types of LOF mutants ( S7 Fig ) might not be relevant for the observed rescued phenotype since they were not normalized in rescue genotyped embryos ., This suggests that other pathways are required for OFT development , and explains , in part , why complete rescue did not occur ., Nonetheless it provides insights as to the nature of the relationship of the two genes , Tbx1 and β-catenin as well as their independent functions ., Loss of β-catenin using various pharyngeal mesoderm engineered Cre drivers , including Mesp1-Cre 38 , Nkx2-5-Cre 39 , Isl1-Cre 40 , 41 and Mef2c-AHF-Cre 27 results in embryonic lethality due to the presence of cardiac outflow tract defects ., The mechanisms mediating these abnormalities , in particularly within the pharyngeal mesoderm of the AHF , have not been well defined ., This is especially important because there are many divergent and distinctive functions of β-catenin during cardiac development 42–45 45 38 , 46 , 47 ., Our interest was to follow up on a previous study in which Tbx1 expression was affected oppositely by loss or constitutively active β-catenin in the pharyngeal mesenchyme using a mesenchymal Cre , termed DermoCre 28 ., Based upon this finding , we investigated the two genes in the AHF tissue at stage E9 . 5 , when the heart tube is elongating ., We found that β-catenin promotes muscle differentiation in the AHF ., We also found that Tbx1 and Wnt/β-catenin act antagonistically to provide a balance of expression of pro-differentiation genes in the AHF that may be required for cardiac outflow tract development ., This sheds new light onto the importance of the two genes in heart development as outlined in the model shown in Fig 7 ., In the model in Fig 7 , we illustrate the Tbx1 expression domain in the SHF as a triangle , with the strongest expression anteriorly , in the AHF tissue and weakest in the posterior SHF ., On the other hand , Wnt/β-catenin expression and function is strongest in the posterior SHF and weakest in the AHF , at E9 . 5 during mouse embryogenesis ., In the panel on the left , we created a simple negative feedback loop , which is consistent with our findings in this study and previous findings using DermoCre 28 ., The center panel of the model illustrates the situation when Tbx1 is inactivated or β-catenin is constitutively active in the AHF ., Here , in these embryos differentiation occurs prematurely in the AHF , prior to reaching the heart tube ( Fig 7 , middle panel ) ., This results in impaired cardiac outflow tract development ., In our study , we found that loss of Tbx1 in the Mef2c-AHF-Cre domain , along with loss of one allele of β-catenin provided significant rescue of heart defects ( Fig 7 , right panel ) ., This supports the importance of their interaction ., However , rescue is not complete ., An explanation for this is that β-catenin is only partially diminished when one allele is inactivated , such that complete normalization is not possible ., Another explanation is that both Tbx1 and Wnt/β-catenin act in many complex pathways in the AHF at this time point ( E9 . 5 ) , for which only the overlapping functions were normalized in rescued embryos 48–51 ., Further , genes changing in expression at E9 . 5 may only be partially reflected in PTA defects observed at E12 . 5 ., We also note that the defects in the cardiac outflow tract between Tbx1 loss and β-catenin gain in the Mef2c-AHF-Cre domain are different 27 , supporting this idea ., In particular , neonatal lethality occurs in Tbx1 LOF embryos due to the presence of a PTA , while gain of β-catenin results in mid-gestational lethality with a short , hypercellular outflow tract ., One of the main functions of Tbx1 in the AHF is to maintain a progenitor state and restrict premature differentiation prior to reaching the elongating heart tube 22 , 24 ., Supporting this idea , constitutive overexpression of Tbx1 in the Mef2c-AHF-Cre domain results in a decrease in expression of muscle differentiation genes 22 ., Based upon our gene expression profiling data , we suggest that Tbx1 may directly or indirectly , repress expression of key transcription factors that regulate this process i . e . , it maintains the AHF cell fate ., We suggest that there is a small decrease in AHF cell numbers but more importantly , a change in cell fate ., Lack of observable morphology defects in the distal pharyngeal apparatus and lack of significant change in proliferation or apoptosis of the AHF progeny at E9 . 5 support this ., Interestingly , in Tbx1 LOF mutant embryos , we found an increase in expression of genes required for cardiomyocyte specification , such as Gata4 , Tbx5 and Smarcd3 ( Baf60c ) 52–55 56–61 35 , 62–65 ., Tbx5 and Gata4 proteins are co-expressed and physically interact to regulate expression of downstream muscle structural protein genes 52 53 ., The combination of Tbx5 , Gata4 and Smarcd3 are sufficient to differentiate mouse embryonic mesoderm to beating cardiomyocytes 54 ., Another intermediate protein is Serum Response Factor ( SRF ) , which directly promotes expression of genes encoding muscle structural proteins that were found increased in Tbx1 null mutant embryos 22 ., Inactivation of Tbx1 resulted in expansion of expression of SRF protein but not mRNA 22 ., Similarly , we did not find Srf expression levels altered in Tbx1 LOF embryos ., Of interest , the above transcription factors may interact with SRF protein to induce differentiation 66 , supporting a continued role of SRF in Tbx1 biology 22 36 , 67 ., Of interest , the Wnt2 , Tbx5 , Tbx20 , Gata4 and Gata6 genes are expressed and have function in the posterior SHF for formation of the inflow tract ., It is not yet known if any are directly or indirectly regulated by Tbx1 ., An additional role of Tbx1 may be to restrict posterior SHF fate in the AHF so as to maintain the appropriate sub-populations within the SHF for proper heart development ., We previously found expression of these posterior SHF genes were greatly expanded in the AHF tissue in Tbx1 global null mutant embryos 24 and were increased in the same tissue in the conditional mutant embryos by qRT-PCR ., It was previously found that Wnt2 and Gata6 act in the same genetic pathway in the posterior SHF during heart development and when inactivated cause atrial septal defects among other anomalies 68 , 69 ., We observed an increase in expression , but did not identify atrial septal defects ., Since we did not observe a severe morphological defect in Tbx1 LOF embryos at E9 . 5 , we suggest that some of these molecular changes will then affect later development ., One of the challenges in human genetics is to identify risk factors of complex traits , such as congenital heart disease 70 , 71 ., The 22q11 . 2DS , although rare in the general population , offers a relatively homogenous cohort to investigate the basis of variable phenotypic heterogeneity among affected individuals ., Rare deleterious DNA variants altered in expression in Tbx1 LOF embryos and acting antagonistically to canonical Wnt/β-catenin , might act as genetic modifiers of CHD in 22q11 . 2DS ., Examination of whole exome sequence of 22q11 . 2DS subjects 72 is underway with a larger cohort , to identify such variants connected to Tbx1 and Wnt/β-catenin gene networks or pathways needed to provide proper balance of critical cell fate choices ., The work here provides a basis in the near future to translate efforts to studies of human subjects ., In this study , we showed that inactivation of Tbx1 in the AHF using Mef2c-AHF-Cre allele , results in a PTA that is also observed in the most seriously affected 22q11 . 2DS ( velo-cardio-facial/DiGeorge syndrome ) patients ., The PTA defect in Tbx1 conditional loss of function mutant embryos , was partially , but significantly rescued by decreasing one allele of the β-catenin gene in this domain , and this also resulted in a normalization of gene expression changes specifically for muscle differentiation but not necessarily for other classes of genes ., Thus , we conclude that Tbx1 in the Mef2c-AHF-Cre domain acts antagonistically with Wnt/β-catenin in the SHF to modulate differentiation prior to entering the heart tube ., The following mouse mutant alleles used in this study have been previously described: Tbx1+/- 7 , Tbx1f/+ ( flox = f ) 73 , Tbx1-GFP 34 , β-cateninf/+ and β-cateninE3/+ 29 , 30 , Mef2c-AHF-Cre/+ 18 , ROSA26-GFPf/+ ( RCE:loxP ) 74 and Wnt/β-catenin signaling reporter mice ( Tg ( TCF/Lef1-HIST1H2BB/EGFP ) 61Hadj/J; TCF/Lef:H2B-GFP 75 ., To generate Mef2c-AHF-Cre/+;Tbx1f/- mutant embryos ( Tbx1 LOF ) , Mef2c-AHF-Cre/+ transgenic male mice were crossed to Tbx1+/- mice to obtain male Mef2c-AHF-Cre/+;Tbx1+/- mice that were then crossed with Tbx1f/f females ., Alternatively , to generate Mef2c-AHF-Cre/+;Tbx1f/f mutant embryos , Mef2c-AHF-Cre/+ transgenic male mice were crossed to Tbx1f/f mice to obtain male Mef2c-AHF-Cre/+;Tbx1f/+ mice , and these were then crossed with Tbx1f/f females ., Wild type and Me2fc-AHF-Cre/+;Tbx1f/+ littermates were used as controls for the experiments ( First Tbx1 LOF and rescue crosses , S1 Table ) ., Tbx1 gain of function embryos ( Tbx1 GOF ) were generated by crossing male Mef2c-AHF-Cre/+ mice with Tbx1-GFPf/f females ., To generate Mef2c-AHF-Cre/+;β-cateninf/f mutant embryos ( β-cat LOF ) , male Mef2c-AHF-Cre/+ transgenic mice were crossed to β-cateninf/f females to obtain male Mef2c-AHF-Cre/+;β-cateninf/+ mice that were then crossed with β-cateninf/f females ., β-catenin gain of function ( β-cat GOF ) embryos i . e , Mef2c-AHF-Cre/+;β-cateninE3/+ or male Mef2c-AHF-Cre/+ transgenic mice were crossed to β-cateninE3/E3 females ., Double mutant embryos were generated by addition of one copy of the β-cateninflox allele to Tbx1 LOF embryos resulting in what we denote as rescue genotyped embryos ., In this case , the females used for the experimental crosses were of the Tbx1f/f;β-catenin f/f genotype , which have been maintained as an inbred line deriving from a mixed C57Bl/6; Swiss Webster background ., The reporter ROSA26-GFP f/+ allele was added to the Tbx1f/f and Tbx1f/f;β-catenin f/f lines when visualizing Mef2c-AHF-Cre lineage ., To evaluate Wnt/β-catenin signaling in wild type embryos , TCF/Lef:H2B-GFP/+ reporter mice were used ., The Mef2c-AHF-Cre/+;Tbx1f/+ and the Mef2c-AHF-Cre/+;Tbx1+/- mice are congenic in Swiss Webster ., The Tbx1f/f;β-cateninf/f mice are in an inbred line , as above ., The Mef2c-AHF-Cre/+;Tbx1+/- x Tbx1f/f; β-cateninf/f crosses were performed 2 years before the Mef2c-AHF-Cre/+;Tbx1f/+ x Tbx1f/f; β-cateninf/f crosses ., The Tbx1f/f and ROSA26-GFPf/+ lines are congenic in Swiss Webster ., The β-cateninE3/+ and β-cateninf/f mice were in a mixed C57Bl/6; Swiss Webster background ., To exclude the possibility that a strain background might affect the possible rescue by β-catenin LOF allele , half of both Tbx1 LOF and the rescue genotyped embryos were generated by using Tbx1f/+;β-cateninf/+ females ( second Tbx1 LOF and rescue crosses; S1 Table ) ., Here , both Tbx1 LOF and the rescue genotyped embryos were littermates ., The PCR strategies for mouse genotyping have been described in the original reports and are available upon request ., All experiments including mice were carried out according to regulatory standards defined by the NIH and the Institute for Animal Studies , Albert Einstein College of Medicine ( https://www . einstein . yu . edu/administration/animal-studies/ ) , IACUC protocol # 2013–0405 ., Institutional Animal Care and Use Committee ( IACUC ) approved this research ., The IACUC number is 20160507 ., Whole-mount RNA in situ hybridization with non-radioactive probes was performed as previously described 76 , 77 , using PCR-based probes , Tbx1 78 , Wnt2 forward primer: 5’ TGGCTCTGGCTCCCTCTGCT 3’ and reverse primer: 5’ CAGGGAGCCTGCCTCTCGGT 3’ and Wnt4 forward primer: 5’ CCGCGAGCAATTGGCTGTACC 3’ and reverse primer: 5’ TGGAACCTGCAGCCACAGCG 3’ ., Following whole-mount protocol , the embryos were fixed overnight in 4% paraformaldehyde ( PFA ) and then dehydrated through a graded ethanol series , embedded in paraffin and sectioned at 10 μm ., Minimum of 5 embryos from 3 independent litters were analyzed per embryonic stage ., After fixation as described above , frozen sections were obtained at a thickness of 10 μm and then permeabilized in 0 . 5% Triton X-100 for 5 min ., Blocking was performed with 5% serum ( goat or donkey ) in PBS/0 . 1% Triton X-100 ( PBT ) for 1 hour ., Primary antibody was diluted in blocking solution ( 1:500 ) and incubated for 1 hour ., Proliferation of cells was assessed by immunofluorescence using the primary antibody anti-phospho Histone H3 ( Ser10 ) , a mitosis marker ( 06–570 Millipore ) ., Sections were washed in PBT and incubated with a secondary antibody for 1 hour ., Secondary antibody was Alexa Fluor 568 goat a-rabbit IgG ( A11011 Invitrogen ) at 1:500 ., Slides were mounted in hard-set mounting medium with DAPI ( Vector Labs H-1500 ) ., Images were captured using a Zeiss Axio Observer microscope ., To perform statistical analysis of cell proliferation , we first counted the Mef2c-AHF-Cre , GFP positive cells in the pharyngeal apparatus located behind the heart in embryo sections and then calculated the average cell counts per tissue section for each embryo ., Then we counted all proliferating cells in each section and calculated the ratio of proliferating cells within the Mef2c-AHF-Cre lineage ., Then , we estimated the mean and standard error of the average cell counts for controls , Tbx1 LOF and rescued embryos and compared them using the t-test ., Apoptosis was assessed on 10 μm thick frozen sections by using TMR Red In situ Cell Death kit ( 2156792 Roche ) following the manufacturer’s instructions ., Natural GFP from the reporter or an antibody for GFP ( Abcam 6290 ) was used to distinguish the AHF cells in both assays described above ., Representations of the complete AHF region from at least 4 embryos per genotype from at least 3 independent litters were used in each assay ., Wnt/β-catenin signaling reporter mice , TCF/Lef:H2B-GFP 75 were used to observe Wnt/β-catenin signaling by direct fluorescence of green fluorescent protein ( GFP ) in wild type embryos at embryonic day E9 . 5 ( 19–21 somite pairs ) ., Mouse embryos were fixed and cryosectioned at 10 μm ., Slides were mounted in hard-set mounting medium with DAPI to visualize DNA ( Vector Labs H-1500 ) ., Images were then captured using a Zeiss Axio Observer microscope ., Nuclear Wnt/β-catenin signaling was counted as the GFP positive signal that co-localized with the DNA ., A minimum of 5 embryos from 3 independent litters was analyzed ., Mouse embryos were isolated in phosphate-buffered saline ( PBS ) and fixed in 10% neutral buffered formalin ( Sigma Corp . ) overnight ., Following fixation , the embryos were dehydrated through a graded ethanol series , embedded in paraffin and sectioned at 5 μm ., All histological sections were stained with hematoxylin and eosin using standard protocols ., Staining was performed in the Einstein Histopathology Core Facility ( http://www . einstein . yu . edu/histopathology/page . aspx ) ., For Tbx1 LOF mutants , a total of 70 hearts at E14 . 5 were obtained from more than 50 independent crosses and analyzed morphologically using light microscopy ., For the rescue crosses , 56 hearts at E14 . 5 were obtained and the Fisher’s exact test was performed to compare the proportion of rescued phenotypes observed between rescued genotype hearts and the Tbx1 LOF mutants ., Images were generated from GFP expressing embryos by direct fluorescence immediately following dissection ., For tissue sections , embryos were fixed for 2 hours with embryos stage ≤ E10 . 5 ( 30–32 somite pairs ) ., Fixation was carried out in 4% PFA in PBS at 4°C ., After fixation , tissue was washed in PBS and then cryoprotected in 30% sucrose in PBS overnight at 4°C ., Embryos were embedded in OCT and cryosectioned at 10 μm ., Images were then captured using a Zeiss Axio Observer microscope ., Embryos at E9 . 5 ( 19–21 somites pairs ) were used for global gene expression studies ., To obtain enough RNA for microarray hybridization experiments , microdissected AHFs ( defined here as: pharyngeal arches 2–6 ) from 27 of each of the following genotypes: Tbx1 LOF and its control ( Tbx1f/+ ) , Tbx1 GOF and its control ( Tbx1-GFP/+ ) , β-cat LOF and its control ( β-cateninf/+ ) , β-cat GOF and its control ( β-catE3/+ ) , rescue and its control ( Tbx1f/+;β-cateninf/+ ) , were pooled in groups of three or six according to the genotype ., For this experiment we used controls that did not have Cre ., Between 4–6 microarrays were performed per genotype in 2–3 batches ., The tissue was homogenized in Buffer RLT ( QIAGEN ) ., Total RNA was isolated with the RNeasy Micro Kit according to the manufacturer’s protocol ., Quality and quantity of total RNA were determined using an Agilent 2100 Bioanalyzer ( Agilent ) and an ND-1000 Spectrophotometer ( NanoDrop ) , respectively ., Biotinylated single-stranded cDNA targets were amplified from 100 nanograms ( ng ) starting total RNA using the Ovation RNA Amplification System V2 and FLOvation cDNA Biotin Module V2 ( NuGEN ) ., A total of 3 . 75 mg of cDNA was hybridized to the GeneChip Test3 array ( Affymetrix ) to test the quality of the labeled target ., Nucleic acid samples that passed quality control were then hybridized to the Affymetrix Mouse GeneST 1 . 0 chip ., Hybridization , washing , staining and scanning were performed in the Genomics Core at Einstein ( https://www . einstein . yu . edu/research/shared-facilities/cores/46/genomics/ ) according to the Affymetrix manual ., Data analysis was performed in the R statistical package ., GeneChip data were pre-processed by the ‘oligo’ package 79 , which implements Robust Multichip Average ( RMA ) algorithm with background correction , quantile normalization and gene level summarization 80 ., Afterwards , for convenience of comparison , only probe-set assigned to genes were kept for subsequent analysis ., Multiple probe-sets for the same genes were collapsed by “average” to obtain a single measurement per gene 81 ., As some arrays were assayed in different batches , we performed UPGMA ( unweighted pair group method with arithmetic mean ) clustering of samples by transcriptomic profile similarities based on the Spearman correlation coefficients ., This analysis indicated clear batch effects , especially for β-cat LOF and Tbx1 LOF data ( data not shown ) ., Hence , we applied ComBat , an efficient batch effect removal approach , to remove batch effec | Introduction, Results, Discussion, Methods | The 22q11 . 2 deletion syndrome ( 22q11 . 2DS; velo-cardio-facial syndrome; DiGeorge syndrome ) is a congenital anomaly disorder in which haploinsufficiency of TBX1 , encoding a T-box transcription factor , is the major candidate for cardiac outflow tract ( OFT ) malformations ., Inactivation of Tbx1 in the anterior heart field ( AHF ) mesoderm in the mouse results in premature expression of pro-differentiation genes and a persistent truncus arteriosus ( PTA ) in which septation does not form between the aorta and pulmonary trunk ., Canonical Wnt/β-catenin has major roles in cardiac OFT development that may act upstream of Tbx1 ., Consistent with an antagonistic relationship , we found the opposite gene expression changes occurred in the AHF in β-catenin loss of function embryos compared to Tbx1 loss of function embryos , providing an opportunity to test for genetic rescue ., When both alleles of Tbx1 and one allele of β-catenin were inactivated in the Mef2c-AHF-Cre domain , 61% of them ( n = 34 ) showed partial or complete rescue of the PTA defect ., Upregulated genes that were oppositely changed in expression in individual mutant embryos were normalized in significantly rescued embryos ., Further , β-catenin was increased in expression when Tbx1 was inactivated , suggesting that there may be a negative feedback loop between canonical Wnt and Tbx1 in the AHF to allow the formation of the OFT ., We suggest that alteration of this balance may contribute to variable expressivity in 22q11 . 2DS . | To understand the genetic relationship between Tbx1 and canonical Wnt/β-catenin , we performed gene expression profiling and genetic rescue experiments ., We found that Tbx1 and β-catenin may provide a negative feedback loop to restrict premature differentiation in the anterior heart field ., This is relevant to understanding the basis of variable expressivity of 22q11 . 2DS , caused by haploinsufficiency of TBX1 . | medicine and health sciences, cardiovascular anatomy, cardiac ventricles, ventricular septal defects, animal models, developmental biology, model organisms, experimental organism systems, embryos, morphogenesis, cardiology, research and analysis methods, embryology, muscle differentiation, birth defects, gene expression, mouse models, congenital disorders, mesoderm, anatomy, congenital heart defects, genetics, biology and life sciences, heart | null |
journal.pcbi.1007327 | 2,019 | Mechanical properties of tubulin intra- and inter-dimer interfaces and their implications for microtubule dynamic instability | αβ-tubulin heterodimers polymerize into microtubules , hollow cylindrical structures , usually composed of 13 laterally attached protofilaments 1 ., Microtubules are about 25 nm wide and range in lengths from tens up to millions of nanometers ., They form cilia and flagella and serve as tracks for long-distance transport of intracellular cargos , such as vesicles and organelles ., In contrast to other polymers , microtubules are highly non-equilibrium systems 2 , which can remain in growth and shrinkage phases with relatively rare spontaneous transitions between them 3 ., Because of this behavior , known as dynamic instability , individual microtubules display significant length changes , even at steady state ., They elongate at their tips by addition of guanosine triphosphate ( GTP ) -bound tubulins ., Soon after incorporation into microtubule lattice , GTP molecules are hydrolyzed to guanosine diphosphate ( GDP ) ., This leads to a conformational change in tubulins , so the lattice made of GDP-tubulins becomes less stable and more prone to depolymerization ., However , because of a lag between the association of GTP-tubulins with microtubules and GTP hydrolysis , there is a certain number of GTP-tubulins at the growing microtubule tip , known as GTP cap , which prevents disassembly until the stabilizing cap is lost 4 ., Both ends of microtubules are dynamically unstable ., The end , exposing β-tubulin subunits , is called the plus-end ., It grows faster than the other end , known as the minus-end ., The origin of the difference of behavior between the plus and minus-ends of microtubules is currently poorly understood ., In cells , the minus-ends are usually capped , so they remain stable ., The plus ends are usually dynamic and serve multiple roles ., During mitosis they generate forces responsible for chromosome motions , leading to segregation of duplicated DNA between daughter cells 5 , 6 ., This fact has been extensively exploited for therapeutics , as the inhibition of microtubules dynamics by small molecule drugs leads to arrest of cell division followed by apoptosis , leading to a powerful method to fight proliferation of actively dividing tumor cells 7 ., Despite extensive studies of dynamic instability for over three decades , the molecular features of GTP- and GDP-tubulins , determining their distinct propensity to polymerize , remain unclear ., Early cryo electron microscopy ( EM ) studies reported very distinct shapes at the ends of growing and shortening microtubules 8 ., That observation informed a so called ‘allosteric’ model of the GTP cap , postulating that GTP hydrolysis induced an allosteric conformational change in straight GTP-tubulin dimers , so GDP-tubulin became curved ., Further cryo EM work modified the allosteric model , proposing that GTP-tubulin was also slightly curved , but still straighter than GDP-tubulin ., The latter modification was based on observations of gently curving extensions on growing microtubule tips 8 , 9 , and the shapes of tubulin structures formed in presence of slowly hydrolysable GTP analogue , GMPCPP 10 , 11 ., Subsequent studies have accumulated substantial evidence indirectly supporting an alternative , ‘lattice’ model of the GTP-cap , postulating that the phosphorylation state of tubulin-bound nucleotide affects the strengths of inter-tubulin bonds , while the shapes of free GTP- and GDP-bound tubulins remain similar ., The following lines of evidence against a significant difference in curvature between GTP- and GDP tubulins have been reported: ( 1 ) free tubulin dimers and tetramers are similarly curved in all available crystal structures of tubulin ( reviewed in 12 ) ; ( 2 ) there is no significant difference in the shape of GTP- and GDP-tubulins according to small-angle X-ray scattering measurements 13; ( 3 ) affinity to allocolchicine , which is thought to be able to bind only a curved intra-dimer interface , is the same for GTP- and GDP-tubulins 13; ( 4 ) recent cryo electron tomography reveals essentially no difference in the curvatures of protofilaments at the tips of growing and shortening microtubules 14 ., Finally , computational studies so far have consistently found that the relaxed tubulin conformation is curved , irrespective of the nucleotide bound , while the strengths of the inter-tubulin bonds are likely to be nucleotide-dependent 15–21 ., We note , however , that at the time when many of those important pioneering simulations were carried out , no high resolution structures of both GTP- and GDP-tubulins were yet available ., The first relatively high resolution cryo-EM-based structures of tubulins in microtubule walls in presence of GDP or GTP analog , GMPCPP , were presented in the seminal paper of Alushin et al . , 2014 22 ., But even those structural data and their subsequent improvements have not yet produced a fully consistent picture that would clearly support only the allosteric or the lattice model ., For example , reports from the Nogales group did not detect any considerable changes at the lateral tubulin-tubulin interfaces and emphasized the “compaction” and skew of GDP- tubulins , in contrast to “extended” GTP-tubulin state in the lattice 22–24 ., The authors proposed that compaction could induce internal mechanical strain in GDP-tubulin ., A study from Moores’ group , however , suggested that lateral bonds were not unchanged , but weakened after GTP hydrolysis 25 ., These controversies , together with the inability of modern structural methods to directly visualize conformational changes of tubulins following GTP hydrolysis and breakage of lateral bonds have encouraged us to undertake a new computational study , in which we re-investigate the effects of nucleotides on the shape and mechanics of tubulins , taking advantage of the wealth of newly available structural data and new computational resources ., A recent cryo electron tomography work indicated importance of tubulin protofilaments , rather than just free dimers , as structural intermediates of microtubule assembly process 14 ., Therefore , both intra-dimer and inter-dimer interfaces could have an important role in dynamic instability ., To take this into account , we carried out molecular dynamics simulations of tubulin protofilaments , which contained both types of inter-tubulin interfaces ., The tubulin protofilaments were extracted from the microtubule wall in compacted , GDP-bound , or extended , GTP-bound , states ., In our simulations , the tubulin protofilaments in each of these nucleotide states relaxed to similar non-radially curved and twisted conformations , in contrast to the expectations of the allosteric model of microtubule instability ., Our further analysis suggested that GTP hydrolysis primarily affected the flexibility and conformation of the inter-dimer interface , without a strong impact on the shape or flexibility of αβ-tubulin heterodimer ., The inter-dimer interfaces of GTP-tubulins were significantly more flexible than those of intra-dimer interfaces ., We argue that such a difference in flexibility could be key for distinct dynamic behavior of plus and minus microtubule ends ., To characterize the nucleotide dependence of tubulin’s shape in relaxed tubulin dimers and short tubulin protofilaments , we prepared all-atom molecular dynamics models of tubulins , extracted from cryo-EM-based structures of GDP- and GTP-like ( GMPCPP ) microtubule lattice 22 ., Flexible tubulin tails were included in the simulation to make sure that their potential effects on tubulin conformations would be taken into account 26 ., For each nucleotide , we carried out two one-microsecond-long simulations of tubulin dimers , and three one-microsecond-long simulations of short protofilaments , representing two longitudinally bonded dimers ( S1 and S2 Movies ) ., In order to put the resulting conformational changes of tubulins into the context of a microtubule , we aligned the α-tubulin subunits of each simulated structure with a microtubule wall fragment , so all types of rotations could be assessed relative to the microtubule-bound coordinate system xyz ( Fig 1 ) ., Consistent with previous reports 15 , 21 , 27 , in our simulations both GDP- and GTP-dimers relaxed over time to similar bent shapes ( Fig 2A ) ., Although the bending occurred predominantly in the ‘outward’ direction , it did not happen in the plane that contained the microtubule axis , as is clearly seen in Fig 2B , which shows projections of the unit orientation vector of the β-tubulin subunit relative to α-tubulin subunit ., Additional molecular dynamics simulations , based on a GDP-tubulin dimer structure extracted from zinc-induced sheets ( PDB code: 1JFF ) , indicated that such non-radial bending and a slight twist were common conformational changes shared by other structures of tubulin ( S1A and S1B Fig ) ., To independently validate the tendency of dimers to curve in non-radial plane , we predicted major low-frequency modes of motion for three tubulin dimer structures , using normal modes analysis ( NMA ) ., The first NM in all types of tubulin dimer structures examined corresponded to predominantly twisting motions of β-tubulin relative to the α-tubulin subunit , while the second and the third NMs represented predominantly non-radial bending , similar to a previous report 28 ., Major low-frequency modes of GTP- and GDP-tubulin dimers were very similar ., Principal component analysis ( PCA ) of molecular dynamics simulation trajectories identified significant correlation of the major NMs with principal components ( PCs ) of motions in molecular dynamics ., The first three PCs explained , on average , 65% of observed tubulin dimer motions ( S1 Table ) ., Overall , the major bending motions were represented by the second NM ( Fig 2C ) ., They occurred in a direction very similar to that of the major conformational changes observed in our molecular dynamics simulations ( Fig 2B ) ., Conformational analysis of tubulin tetramers revealed that additional inter-dimer interface did not qualitatively change the overall fashion of bending of the whole tetramer , compared to that of a tubulin dimer ., Specifically , both GDP- and GTP-tubulin tetramers also assumed outwardly curved shapes in the end of one-microsecond-long simulations ( Fig 2D and Fig 2E ) ., The overall bending was also non-radial ., Likewise , the dominant NM of the whole tetramer corresponded to bending in a similar non-radial direction ( Fig 2F , S3 Movie ) and it overlapped significantly with PCs of motions in molecular dynamics ( S2 Table ) ., The first three PCs explained , on average , 79% of the total variance present in the molecular dynamics trajectories of tetramers ( S2 Fig ) ., To gain more detailed insight into the conformational changes at intra- and inter-dimer tubulin interfaces during their relaxation from straight to curved shapes and their dependence on the associated nucleotide , we described the relative motions of adjacent tubulin monomers at each interface with three rotation angles , using metrics similar to those introduced previously 16 ., Specifically , bending at the tubulin-tubulin interface was described with two angles: θ-angle characterized the magnitude of the conformational change , while auxiliary angle , φ , showed the direction in which the bending occurred; δ-angle characterized twist of one tubulin monomer relative to the other ( see Methods for more details ) ., To determine the direction of rotations relative to the microtubule structure , we aligned the minus-end-proximal tubulin monomer onto a corresponding subunit in a straight microtubule fragment , which was oriented relative the coordinate system as depicted in Fig 1 ., In this arrangement , outward strictly radial bending is described by positive θ-angles and φ = 0 degrees ., Despite the advantage of being physically clear and easy to relate with microtubule geometry and with the degrees of freedom , which are usually present in higher-scale models of microtubule dynamics , the rotation angles may not optimally represent the multi-dimensional molecular dynamics data ., For this reason , we also carried out an additional analysis , projecting molecular dynamics trajectories on two main PCA modes and comparing the movements at tubulin interfaces , expressed in those observables ( S3 Fig , S4 Fig , S4 Movie ) ., This yielded essentially similar conclusions about the relative properties of the inter- and intra-dimer interfaces , so we decided to stick to rotation angles throughout this report for the sake of intuitiveness and physical clarity of description ., First , we calculated rotation angles in tubulin tetramers at intra-dimer tubulin interfaces ., After 500 ns of simulation , intra-dimer interfaces of both GDP- and GTP-tubulin tetramers relaxed to a conformation , in which their β-subunits were tilted relative to α-tubulins with almost identical magnitudes of intra-dimer bending , 9 . 4 ± 0 . 9 and 8 . 2 ± 0 . 7 degrees , respectively ( Fig 3 , Table 1 ) ., Simultaneously , β-tubulins twisted relative to α-tubulins by about 5 . 2 ± 0 . 9 and 7 . 0 ± 1 . 5 degrees , in GDP- and GTP- states ., Conformational changes of free dimers at the intra-dimer interface were quantitatively similar ( Table 1 , S5 Fig ) ., In contrast to intra-dimer interfaces , the inter-dimer interfaces were less reproducible in their bending directions from run to run , suggesting the presence of multiple local minima in the energy landscape ( Table 1 , Fig 4 ) ., Within each run , the bending angle had a satisfactory convergence to a stable mean value , as assessed by splitting the last 500 ns of the simulations in four 125-ns-long segments and analyzing them separately ( S7 Fig ) ., We compared equilibrium conformational angles of tubulins at the end of individual simulation runs with the respective intra- and inter-dimer angles calculated for published crystal structures of tubulins bound to microtubule associated proteins ( MAPs ) : stathmin , darpin , TOG-domain or MCAK proteins 29–34 ., Intra-dimer curvature , its direction and magnitude of twist in simulated structures were similar to characteristics of the experimental structures , confirming the ability of molecular dynamics simulations to predict equilibrium shapes of tubulin dimers ( Table 2 , S1C Fig ) ., Inter-dimer angles , though , were much more variable in simulations , without clear correlation with the corresponding angles in crystallized tubulin-MAP complexes ., We speculate that in complex with a MAP , the tubulin tetramer is likely to be fixed by its interaction partner , resulting in a different direction of inter-dimer bending and in lower flexibility of the inter-dimer interface ., Hence , the conformational variability is reduced ., Despite the fact that in our simulations and in experimental structural data tubulin oligomers both display non-radially curved and twisted shapes , recent cryo electron tomography studies reported nearly planar protofilaments at the tips of growing and shortening microtubules 14 , 35 ., Puzzled by this discrepancy , we performed additional molecular dynamics simulations of GTP- and GDP-tubulin hexamers , applying position restrain on Cα atoms of the minus-end proximal tubulin subunit ( Fig 5A , S4 Movie ) ., Fixation of the terminal α-tubulin was mimicking oligomer attachment to the microtubule end ., Analysis of the relaxed shapes of GTP- and GDP-tubulins after 500 ns simulation revealed that the interfaces proximal to the fixed subunit ( #1–3 ) tended to be overall straighter , less flexible , and bent in a more radial direction ( Fig 5B ) ., On the other hand , distal interfaces ( #4 and #5 ) behaved essentially like those of free tubulin oligomers ( compare with Fig 4C , Fig 4D , Fig 3C and Fig 3D ) ., Given this ‘straightening’ effect due to the longitudinal attachment to the plus tip , overall bending direction of whole hexamers , characterized by projections of the center of mass of β-tubulin subunit onto XY-plane , was somewhat more radial on the scale of hexamers , compared with free tetramers ( Fig 5A , top view ) ., Although , we note that at the scale of longer protofilaments , the tangential component may still be significant ., We also hypothesized that the presence of adjacent protofilament neighbors could affect the bending direction of a given protofilament , attached to the microtubule tip ., To test that , we constructed a molecular model of three laterally bound GDP-tubulin hexamers , whose minus-end proximal α-tubulins were fixed ( Fig 6A , S5 Movie ) ., In two independent simulations of this system , the whole assembly of three protofilaments consistently bent asymmetrically: the splaying amplitude of the right protofilament was the highest , while the left protofilament remained almost straight ( ‘left’ and ‘right’ are defined as in Fig 6A and Fig 6B ) ., This result can be explained by the tendency of individual protofilaments to bend and twist in the directions depicted in Fig 6B , Fig 2D and Fig 2E ., Such motions tend to stretch the left lateral bond significantly less , compared to the right bond ., Therefore , it is the right lateral bond , which restricts the motion more significantly ., Hence , in the absence of the right lateral bond the right protofilament undergoes relaxation to the bent and twisted state relatively easily , while the left protofilament is almost fully restricted by its right lateral bond ( Fig 6B and Fig 6C ) ., As a result , the splaying amplitude of the right protofilament is a lot more dramatic ., This splaying leads to breakage of the lateral bond between the terminal β-tubulin subunits of the right and the middle protofilaments during the simulation , while the bond between the terminal β-tubulin subunits of left and the middle protofilaments remained intact , as we verified by counting the number of contacts between amino acids of those subunits ( Fig 6D ) ., Comparison of protofilament bending in the simulations in the presence or absence of adjacent protofilaments did not reveal a marked effect of lateral neighbors on the bending direction of the middle protofilament ( S8 Fig ) ., We noticed that the variance of rotation angles at the inter-dimer GTP-interface was higher than at all other types of interfaces , which can be appreciated by the highest scatter of projections of the unit OZ-vector , characterizing direction and amplitude of tubulin bending ( Fig 4D vs . Fig 4C , Fig 3C , Fig 3D ) ., Similar increased variance is seen at the inter-dimer GTP-interface , when the data are examined by projecting on the two main PCA modes ( S3 Fig , S4 Fig , S4 Table , S5 Table ) ., Guided by this observation , we hypothesized that the nucleotide could have a distinct effect on the mechanical properties of the inter-dimer interface ., We therefore used two methods to quantify the flexural stiffness of tubulin interfaces ., The first method was based on the equipartition theorem ., Assuming that tubulin structures were already equilibrated by 500 ns of simulation , we pooled tubulin angles after that time and calculated their variance ., According to equipartition theorem , at thermodynamic equilibrium variance , σ2 , of a given conformational angle ( θ or δ ) should be inversely proportional to the respective harmonic flexural stiffness κ:, κ=kBTσ2, ( 1 ), where kB is the Boltzmann constant , T is the temperature ., Resulting harmonic stiffness values are summarized in Table 3 ., These results suggest that the inter-dimer interface of GTP-tubulin is considerably more flexible than its intra-dimer interface in all kinds of rotation ., Intra-dimer stiffnesses are not sensitive to nucleotide , speaking against the presence of significant allosteric effects of GTP hydrolysis on mechanical properties of the intra-dimer interfaces at equilibrium ., Inter-dimer stiffnesses , however , are significantly lower in the GTP-tubulin model compared to the GDP-tubulin ., The latter finding is suggestive of at least a partial contribution of flexural stiffness modulation from the nucleotide hydrolysis state into the mechanism of dynamic instability ., The equipartition theorem-based method has allowed comparing stiffnesses , corresponding to motions in a relatively narrow high-frequency range ., To identify stiffnesses , characterizing conformational changes at longer timescales , we used NMA as a complementary approach ., The squared mode frequency , related to each normal mode , can be reckoned into mechanical properties , such as bending stiffness/torsional rigidity , corresponding to the motion along the given mode 36–38 ., As illustrated by Table 4 , NMA confirmed that inter-dimer interface was much more flexible than the intra-dimer interfaces in the GTP-state , but that was not true for the inter-dimer interface of GDP-tubulin tetramer ., It is tempting to hypothesize that significant stiffening of the inter-dimer interface may be caused by inter-dimer compaction of the GDP-tubulin ., But such compaction could well be released upon relaxation of GDP-tubulin tetramer , extracted from microtubule lattice ., So we questioned , whether or not free GTP- and GDP-tubulin tetramers converged in the simulations to conformations with similar extent of the inter-dimer compaction ., For straight microtubule lattice , intra-dimer and inter-dimer distances were previously used to characterize the extent of tubulin compaction ., They were defined as the lengths of the vectors , connecting ribose rings of the nucleotides in the corresponding pairs of longitudinally bonded tubulins 23 ., Applying the same metric for our simulated tubulin tetramers and averaging over the second half of one-microsecond-long simulations , we found that both GTP- and GDP-tubulins displayed similar extended intra-dimer and similar shorter inter-dimers distances ( S3 Table ) ., Interestingly , these numbers closely matched the corresponding distances in crystal structures of curved tubulins , e . g . in the structure of tubulin tetramer in complex with stathmin and vinblastine ( S3 Table , 30 ) ., However , we note that this metric should be used with caution for describing compaction of curved tubulin structures , because in this case , the inter-tubulin interfaces may not shrink or extend predominantly along the vector , connecting the nucleotides in adjacent tubulins ., Moreover , other substantial conformational changes may be present ., Therefore , we decided to additionally characterize the inter- and intra-dimer interfaces with the number of contacts between α- and β-tubulin amino acids at the interface ., In fact , the number of contacts should correlate with compaction because amino acids at the more compact interface come closer together ., We find that the inter-dimer interface of GDP-tubulin retains the largest number of contacts throughout the simulation ( S6 Table ) ., Thus , the high number of contacts at the GDP-tubulin inter-dimer interface might explain the enhanced flexural stiffness of this interface ., Taking advantage of the new generation of cryo-EM-based in-microtubule tubulin structures in GDP and GTP-like states , we carried out several one-microsecond-long molecular dynamics simulations of free tubulin dimers and tetramers ., Our simulations reveal that initially “compacted” GDP-bound tubulins , and initially “extended” GTP-bound tubulins both adopt similar non-radially curved and slightly twisted shapes ., The presence of substantial tangential bending and twist components in the resulting relaxed tubulin conformations , is fully consistent with published crystal structures of tubulins in complex with MAPs 29–34 ( Tables 1 and 2 ) ., However , in contrast to those data , no pronounced out-of-plane bending was observed in a recent cryo electron tomography study , which reported essentially flat protofilaments , lying mainly in the radial planes , containing the microtubule axis 14 , 35 ., The origin of this discrepancy is not completely clear ., But it could be partially explained by the conformational effects , induced by attachment of tubulin oligomers to the plus-end of the microtubule , as suggested by our simulations of single and three tubulin hexamers , longitudinally fixed at the minus-end ., It might also be possible that several longitudinally attached tubulins experience some kind of cooperative behavior , similar to described with FtsZ 39 ., Strikingly , we find that the nucleotide type does not affect the curvature or mechanics of the intra-dimer interface , but it does considerably modify the stiffness of the inter-dimer interface ., Prior to GTP hydrolysis the inter-dimer interface is significantly softer than the intra-dimer one ., Based on our simulations , we propose that enlarged number of contacts at the inter-dimer interface in the GDP-state makes it even slightly stiffer than the intra-dimer interface ., In our opinion , these findings have at least three important implications for our understanding of the mechanisms of microtubule instability ., First , non-radial bending and twisting of tubulins during their relaxation from straight to curved configuration likely means that the lateral bonds on two sides of the splaying protofilaments at the microtubule tip experience unequal mechanical stress ., This conclusion is qualitatively similar to the results of a previously study , which considered a hypothetical microtubule tip , constructed based on X-ray structures of curved tubulins in complex with stathmin 40 ., We speculate that uneven distribution of mechanical stress on lateral bonds may lead to sequential , rather than simultaneous , rupture of the lateral bonds , which has not been previously considered by the majority of existing models of microtubule dynamics 41–46 , with one notable exception 45 ., This proposal is illustrated by our simulations of three laterally attached protofilaments , which splay apart by breaking lateral bonds between the right and the middle protofilaments sooner than the bonds between the middle and the left protofilaments ( Fig 6 ) ., Given the fact that sequential breakage of the lateral bonds is energetically more feasible than their simultaneous breakage , we believe that taking this new feature of tubulin mechanics into account may lead to significant revision of current estimates for lateral bonds between tubulins ., More work is needed to investigate full lateral bond rupture and its dependence on the nucleotide ., Second , stiffening of the compacted interface between GDP-tubulin dimers helps to resolve the apparent paradox of a lack of clearly visible effects of the nucleotide on lateral bonds and on the curvature of free tubulin protofilaments , despite a well-established link between microtubule stability and the type of nucleotides associated with the microtubule lattice ., Indeed , most experimental work and theoretical thinking in the literature have been focused on an effort to explain dynamic instability by nucleotide-dependent curvature or by nucleotide-dependent strengths of lateral bonds ., The possibility of modulating tubulin flexural stiffness by the nucleotide has been raised by only a small number of studies , e . g . 10 , 21 , 27 , and underappreciated ., Recently Igaev and Grubmüller suggested an allosteric mechanism for dynamic instability , based on the nucleotide-driven tubulin dimer stiffness change 21 ., Our current study essentially points to a very similar idea: the softer interface between GTP-tubulin dimers requires less additional energy to straighten curved GTP-tubulin protofilaments in order to incorporate them into the microtubule wall , compared to a stiffer GDP-tubulin protofilament ., However , importantly , in this study we did not observe any significant difference in flexural stiffnesses between intra-dimer interfaces of GTP and GDP-tubulins ., Instead , the nucleotide dramatically affected the inter-dimer interface around the exchangeable nucleotide binding site ., This inter-dimer interface was not examined in the former study , which was focused on free tubulin dimers ., We do not think that the disagreement in our conclusions about tubulin dimers could be related to the tubulin structures , or molecular dynamics simulation parameters we used , as they were similar ., But there were differences in the approach , with which the flexibilities were assessed ., Here we measured the flexibility of tubulin interfaces around their relaxed curved conformations , while Igaev and Grubmüller used the umbrella sampling method to probe energetic landscape on a larger scale along a reaction coordinate , which did not exactly correspond to the bend and twist angles , which we used in this study ., An equipartition-based analysis , similar to ours , was first carried out in the pioneering paper by Grafmüller and Voth 15 ., In contrast to the present study , the authors did not find any statistically significant nucleotide-dependence of either intra- or inter-tubulin interfaces ., The stiffness of inter-dimer interfaces clearly depended on the structures they used ., At the time of the study , only a low resolution structure of straight GDP-bound tubulin from Zn-induced sheets with antiparallel protofilaments was available ( PDB code: 1JFF ) ., That tubulin structure was compacted around the exchangeable nucleotide-binding site , so it is unclear if simple insertion of Mg2+ ion and substituting GDP molecule for GTP would drive a correct GTP-tubulin conformation , given the complexity of tubulin’s energy landscape ., The authors also examined the relaxation of tubulins from the curved structure ( PDB code: 1SA0 ) , which represented a complex of colchicine , stathmin and GDP-tubulin tetramer ., Although the structure had a higher resolution , the interface between dimers in presence of stathmin could be affected by this MAP ., Moreover , computational resources now allow substantially longer simulation times , which increase the chances of more complete relaxation of the tubulin structure ., Finally , we find unequal flexural stiffness of the inter-dimer and intra-dimer interfaces of GTP-tubulins , and less different but still distinct flexural stiffness of inter- and intra-dimer interfaces of GDP-tubulins ., We argue that these features could be essential for explaining the difference in the rates of assembly and disassembly of plus- and minus-ends of microtubules ., It has recently been proposed , that microtubules assemble and disassemble by dynamic peeling and unpeeling of curved GTP-protofilaments at the microtubule tip , so that the balance between the lateral bonds and outward bending/twisting energy controls the rate of both microtubule assembly and disassembly 14 ., The minus-ends of microtubules terminate with α-tubulins , while the plus-ends terminate with β-tubulins ., Obviously , if the lateral bonds between α-α and β-β tubulins were equal , microtubule rates would be identical at both ends ., Distinct lateral bonds between α-α and β-β tubulins alone cannot render the rates different either , if the flexural stiffnesses between any pair of tubulin subunits is the same ., However , if the inter-dimer interface is soft , compared to the intra-dimer interface , the terminal layer of subunits is under higher bending stress compared to the layer second to terminal ( Fig 7 ) ., Indeed , the lateral bonds of the terminal layer are opposed by the large bending force , coming from stiff intra-dimer interface , trying to curve the subunit out and break the lateral bonds , so these layers should break lateral bonds relatively fast ., The layer second to terminal , on the other hand , is under lower bending stress , because the inter-dimer stiffness is significantly softer ., Therefore , lateral bonds holding the second to terminal layers would exist for longer time and become rate limiting ., This means that the second to terminal layer , composed of β-tubulins at the minus-end and α-tubulins at the plus-end can control the rates of assembly from GTP-tubulin , explaining why those rates could differ ., An alternative but not mutually exclusive explanation of the difference in plus- and minus-end assembly/disassembly rates , obviously , is a poss | Introduction, Results, Discussion, Methods | Thirteen tubulin protofilaments , made of αβ-tubulin heterodimers , interact laterally to produce cytoskeletal microtubules ., Microtubules exhibit the striking property of dynamic instability , manifested in their intermittent growth and shrinkage at both ends ., This behavior is key to many cellular processes , such as cell division , migration , maintenance of cell shape , etc ., Although assembly and disassembly of microtubules is known to be linked to hydrolysis of a guanosine triphosphate molecule in the pocket of β-tubulin , detailed mechanistic understanding of corresponding conformational changes is still lacking ., Here we take advantage of the recent generation of in-microtubule structures of tubulin to examine the properties of protofilaments , which serve as important microtubule assembly and disassembly intermediates ., We find that initially straight tubulin protofilaments , relax to similar non-radially curved and slightly twisted conformations ., Our analysis further suggests that guanosine triphosphate hydrolysis primarily affects the flexibility and conformation of the inter-dimer interface , without a strong impact on the shape or flexibility of αβ-heterodimer ., Inter-dimer interfaces are significantly more flexible compared to intra-dimer interfaces ., We argue that such a difference in flexibility could be key for distinct stability of the plus and minus microtubule ends ., The higher flexibility of the inter-dimer interface may have implications for development of pulling force by curving tubulin protofilaments during microtubule disassembly , a process of major importance for chromosome motions in mitosis . | The ability to self-assemble from tubulin dimers in presence of guanosine triphosphate ( GTP ) and spontaneously disassemble , when GTP molecules in tubulin pockets are hydrolyzed , is a dramatic and essential feature of microtubules ., This behavior has many important roles , including chromosome segregation in mitosis , rapid remodeling of the microtubule networks , establishing cell polarity and other ., Nevertheless , the mechanism , linking the associated nucleotide with the conformational changes in tubulins , remains elusive ., Most studies suggested that the nucleotide should affect either the equilibrium shape of tubulin dimers or the strengths of the lateral bonds between them ., But existing experimental methods have lacked spatio-temporal resolution to test that ., Theoretical studies , until recently , have suffered from the absence of high-resolution microtubule structures with different nucleotides to build on and the lack of computational efficiency to examine large tubulin assemblies ., Here we use recent cryo electron microscopy structures of GDP and GTP-like microtubules , and employ all-atom molecular dynamics simulations to examine tubulin protofilaments ., We find that the nucleotide primarily affects the interface between two tubulin dimers , making it more flexible in the GTP state ., This makes the GTP-bound tubulin protofilament easier to incorporate into microtubule lattice , providing a simple mechanism for microtubule dynamic instability . | stiffness, mechanical properties, molecular dynamics, microtubules, classical mechanics, statistics, multivariate analysis, mathematics, materials science, damage mechanics, tubulins, oligomers, cellular structures and organelles, cytoskeleton, research and analysis methods, bending, polymer chemistry, proteins, mathematical and statistical techniques, deformation, principal component analysis, dimers, chemistry, statistical methods, physics, biochemistry, cytoskeletal proteins, biochemical simulations, cell biology, biology and life sciences, physical sciences, computational chemistry, materials, material properties, computational biology | null |
journal.pcbi.1003369 | 2,013 | Protein-Protein Interactions in a Crowded Environment: An Analysis via Cross-Docking Simulations and Evolutionary Information | Protein-protein interactions ( PPI ) are at the heart of the molecular processes governing life and constitute an increasingly important target for drug design 1–4 ., Given their importance , it is clearly vital to characterize PPIs and notably to determine which protein interactions are likely to be stable enough to have functional relevance ., Computational methods such as molecular docking have rendered possible to successfully predict the conformation of protein-protein complexes when no major conformational rearrangement occurs during the assembly 5–11 ., However , we 12 and others 13 , 14 have demonstrated that docking algorithms are unable to predict binding affinities and thus , presently , cannot distinguish which proteins will actually interact ., This leads to ask whether this failure comes from the fact that scoring functions , used to sort the docking solutions , are inefficient for partner identification or whether the difficulty comes from binding promiscuity between proteins in the cell that blurs the interaction signal of the functional partners ., In the crowded cell , proteins experience non-specific and unintended interactions with the intracellular environment leading to a severe competition between functional and non-functional partners 15–19 ., This brings to light the importance of characterizing weak , potentially non-functional , interactions in order to predict functional ones and understand how proteins behave within a crowded environment 16 , 20 , 21 ., In this work , we tackle two distinct but related questions:, ( i ) can a combination of coarse-grain docking and evolutionary information identify true interacting partners among a set of potential ones ?, ( ii ) what is the effect of binding promiscuity on a large and variate dataset of protein structures 22 ?, Previously , we have shown that knowing the experimental binding site of a protein can help to retrieve its native interacting partner within a set of decoys 12 ., On the other hand , recent studies reveal that arbitrary docked partners bind in a non-random mode on protein surfaces 23 , 24 suggesting that docking true but also false partners can help to identify protein binding sites ., We developed a novel score based on arbitrary docking and evolutionary information to predict protein binding sites ., The different docking conformations of a given protein pair are scored according to their associated energy and the agreement between the docked interface and the predicted binding sites ., An interaction index is defined , and normalized according to the whole set of proteins tested , in order to discriminate the interacting partners from the set of tested interactions ., We evaluate our method with a complete cross-docking ( CC-D ) calculation on a set of 168 proteins belonging to the 84 known complexes described in the Mintseris Benchmark 2 . 0 25 and covering a large spectrum of different protein interfaces ., Enzymes , inhibitors , antibodies , antigens , signaling proteins and others have been considered as well as interfaces that do or do not undergo conformational adjustments during interaction ., Docking calculations are made with no knowledge of the experimental complex structure: unbound structures are used ., We use a coarse-grain docking algorithm 12 , whose energy function relies on both van der Waals and Coulomb potentials ., We show that the combination of a coarse-grain docking algorithm with binding sites prediction can significantly contribute to the identification of a reasonably sized set of potentially interacting proteins , that can be further investigated by more precise docking algorithms or laboratory experiments ., The large computational effort necessary to accomplish this work was realized with the help of World Community Grid ( WCG ) , that coordinated thousands of internautes providing their computer time to dock about 300000 conformations per protein pair for the set of 28224 possible pairs in the Mintseris Benchmark 2 . 0 ., For each pair , we selected about 2000 decoys ., For non-partners , we find weak as well as strong interactions ., The decoy set is released and it provides an important reference set of structures that can serve as a proxy for the non-specific protein-protein complexes that occur transiently in the cell or that are avoided by spatial-temporal constraints ., These latter are hard to characterize experimentally but they are of biochemical relevance , as highlighted by other studies 26–29 ., To simulate the variability of crowded environments for a protein in the cell , we study how easily a protein finds its true partner with respect to many random subsets of proteins supposedly competing with it ., We realize a thorough analysis of these interactions and we address the question of whether a successful prediction of a protein partner depends on the environment composition or not ., We quantify the effect of competing partners in predictions , and we characterize in a quantitative manner three distinguished populations of proteins interacting with a protein : those that strongly compete with the true partner of , those that never compete with it , and those that interact with with variable levels of strength ., For each protein , we propose a numerical index that provides the strength of the interaction with all other proteins in the environment , and that gives a signature for ., To our knowledge , this is the first study performing a large-scale CC-D calculation , proposing an analysis of the binding promiscuity of the protein set , and providing to the scientific community the associated dataset of decoys 30 , 31 at the same time ., Previous large-scale analyses used docking by shape complementarity that quickly scans through several thousands proteins in a matter of seconds 32 , 33 but ignore the electrostatic contribution playing however an important role in protein interactions 34–37 ., We compared our method to two previously done studies 32 , 33 ., Both of them do not perform a CC-D experiment , but a large-scale analysis of selected protein pairs ., Finally , we checked whether evolutionary information can be used to considerably restrict the number of docking interfaces to be examined and to render molecular computation feasible for a larger scale investigation of PPIs , based on thousands of proteins instead of hundreds ., This result makes the protocol proposed here feasible for scaling up the analysis ., For each protein in the dataset , the problem of partner identification is tackled with two main experiments ., The first experiment assumes to know the residues belonging to the experimental interface of the proteins ., This means that the residues lying at the interface of two proteins in a native complex are supposed to be known while no knowledge of the complex conformation is assumed ., The second experiment replaces experimental interfaces by predictions of binding sites based on docking and evolutionary information ., The evaluation of the quality of the interaction signal in this PPI large-scale study is of major importance ., In particular , the contribution of docking information compared to evolutionary information in partner identification needs to be quantified ., To do so , the analysis based on experimental interfaces allows us to evaluate in a precise manner how much a good prediction of the interaction sites improves partners identification , experimental interfaces playing the role of perfect predictions ., In the sequel , we also use it to decipher whether a property of protein interactions that has been observed from computational predictions has a biological origin or whether it is a consequence of the noise of the prediction ., We performed a series of tests to check whether the composition of a set of competing partners for a given protein influences partnership prediction ., The analysis is performed on both JET+NIP predictions and experimental interfaces ( see Figures S10–S16 , S17–S23 and Table S1 in Text S1 ) ., A few large-scale studies that wish to identify true interacting partners among a set of potential ones , have been recently proposed ., They are computationally demanding and they remain , for this reason , rare ., All large-scale studies we compared to have been based on shape complementarity to quickly scan through several thousand ligands in a matter of seconds ., These approaches do not include any electrostatic component in their energy model , while electrostatic forces are known to play an important role in PPI ., Notice that , given a protein , no other docking studies besides this one tries to quantify the effect of binding promiscuity of a large and variate dataset of protein structures interacting with ., The docking technique we used is computationally expensive ( see “Computational implementation and data analysis” in Methods ) ., To reduce the conformation space to be explored , we predicted the location where the interaction takes place and confined the docking to this region ., This is done by predicting binding sites for the receptor protein by using JET 38 and by defining an appropriate cone around the predicted interface ( see Methods and Figures S5 , S6 in Text S1 ) ., When restricting the docking conformational space with JET , we observe a slight decrease of the AUC ., By using experimental data , the AUC goes from 0 . 84 to 0 . 80 while using predictions , it goes from 0 . 61 to 0 . 59 ( Table 4 ) , revealing a reduced loss in precision ., This shows that using evolutionary information from sequences is a very promising approach to reduce docking computational time ., To evaluate the impact of our restriction on MAXDo execution time , we computed how many docked conformations between protein pairs were dropped ., When the 168 proteins are considered together , the average portion of the conformational space that is explored after reduction is of the original space ., This value should be understood at the light of protein sizes , as illustrated in Figure S59 in Text S1 ., In fact , small proteins require to explore about of their original conformational space , while for large ones , the space is reduced to of the initial one ., This is because small proteins are rather conserved and JET predicts large patches as their interaction sites , covering a large portion of their protein surface ., Notice that this calculation takes into account a reduced number of conformations for the receptor , independently on whether the conformational space of the ligand is completely explored or not ., Clearly , the actual computational time depends on the number of conformations that are tested , and if both the conformational spaces of the receptor and of the ligand are reduced , the effect will be quadratic ., The small difference in AUC obtained by exploring the reduced space of the receptor compared to the whole ( with a fully explored surface of the ligand ) , is due to the high specificity of JET and to the definition of the cone ( see Methods ) that takes into account JETs lower PPV ., We have addressed the problem of predicting protein interactions using high-throughput CC-D calculations on a dataset of 168 proteins ., We have shown that a simple docking algorithm combined with evolutionary information , can be used to discriminate interacting from non-interacting proteins ., The purpose of the method is the in silico large-scale screening of protein structures to find a small set of potential protein partners that could be tested experimentally ., The approach reminds the one of drug design aiming to screen large sets of small molecules in order to identify a small set of potential drugs that becomes experimentally testable ., These approaches do not pretend to exactly identify a unique solution but rather a set of reasonable candidates , and reduce , in this manner , the amount of experimental time and costs ., This means that we are not focused on the correct docking of experimentally known partners , which can be achieved via other more effective but much more computationally demanding methods 43 ., However , one can envisage to use such more sophisticated methods on the small set of candidates that our coarse method identifies to propose more precise models of the potential complexes ., We have realized a large-scale PPI analysis by assuming to know the residues forming the experimental interface of the native complexes ( no associated experimental conformation is considered ) and by using predictions of binding sites ., Experimental binding sites can be seen as perfect predictions , and the analysis based on them is realized for two reasons:, 1 . to understand how much evolutionary information can contribute to PPI reconstruction when coupled with a coarse-grain docking algorithm using an energy function , and, 2 . to decouple true PPI signal from noise and identify PPI properties that are not consequences of accumulated errors due to predictive algorithms ., This second reason allowed us to be confident , for instance , on the promiscuity observed in Figure 5B ( bottom black dots ) by ensuring that it is not generated by noise in predictions ( see Figure 5A ) ., A few large-scale analyses , that are similar in spirit , have been performed 32 , 33 ., A comparison of our results with 33 , based on the ten protein complexes discussed in detail in 33 , reveals a similar performance of the two methods ., However , a full comparison with 33 is impossible since they treat only a subset of the Mintseris dataset , use a large background set and do not provide a detailed measure of the performance of their method ., On the contrary , our method is tested on all complexes of the Mintseris dataset , a good testing platform for methods dedicated to protein partner prediction due to its numerous structural differences ., The global analysis of the two methods ( over the subset of 56 proteins; see Table 1 and 33 ) highlights that we can reasonably search for protein partners within sets of a few hundred monomers ., We demonstrated that improving current predictive methods is possible through a better prediction of binding sites , and we precisely estimated the effect of such predictions ., We could only partially compare to 32 since they do not perform a CC-D of the Mintseris dataset but only cross the 84 receptors against the 84 ligands , that is a fourth of the interactions explored in our analysis ., Performances of our method and the one reported in 32 are comparable on the common subset , but notice that contrary to 32 , we use unbound structures and we make no use of the non-naive split of the dataset ( that is , receptors versus ligands ) ., The predictive performance of the method is encouraging for the whole Mintseris Benchmark 2 . 0 and very satisfactory for the enzyme-inhibitor subset ( Table 2 ) ., For this latter , the AUC reaches a very high value of while the AUC for the whole Mintseris dataset is ., Notice that the way we computed the AUC is very strict , since we asked the true partner to be ranked first over the tested dataset ., A more relaxed evaluation is reported in Table 1 where we show that a fourth of the 168 proteins in the Mintseris dataset are recognized by looking at the top 17 predictions over the 168 tested partners ., If the binding site of the proteins is correctly predicted , the half of the proteins in the dataset are recognized by looking at the top 8 predictions , and two third by looking at the top 17 ., This is a very encouraging result with respect to the potential applicability of this in silico predictive approach to the reconstruction of PPI networks ., In fact , proposing to a biologist a set of less than 20 interactions to test is very reasonable ., The analysis on the average IR for the enzyme-inhibitor subset highlights that an average IR threshold allows the method to propose about 12 partners , a reasonable number of proteins to be selected for experimental tests ., In 38 cases over 46 ( Figure 6 ) , the true partner is present in the retained subset showing a very high sensitivity ., For the whole Mintseris benchmark , for roughly the half of the dataset ( 82 proteins ) , the true partner is retrieved with an average IR ., Notice that when considering the experimental binding site of each partner , 138 proteins over 168 display an average IR ., This means that a precise binding site prediction method will lead to a successful partners discrimination , a problem that could be considered as being much more ambitious than the binding site prediction problem ., Again , these results support the feasibility of the approach to identify potential partners but , most of all , they highlight the interest of testing a protein within a large environment , by randomly choosing many small subsets of proteins in the environment , and by selecting as potential partners to be experimentally tested , those proteins that present a stable average IR ( black dots , Figure 5 ) with the protein under study ., The selection of 10 potential partners instead of 17 ( as suggested by the direct evaluation of the NII matrix in Figure 1 and Table 1 ) might be crucial for experimental validation ., This observation opens a way to new computational schema for partner predictions ., The analysis highlights an important point on the behavior of all proteins with respect to their partners ., For each protein , there is a small set of partners that displays a systematic ( black points in the bottom of Figure 5AB ) very low average IR that lead to ask whether these partners might physically interact and not be false positives ., Three reasonable explanations for this set of highly potential partners can be given:, ( i ) partners can interact on a merely physical base but never meet in the cell due to different cellular compartments localization ,, ( ii ) partners can interact for functional purposes , possibly not described until now ( several different partners are expected to interact with a protein ) ,, ( iii ) partners can interact in the cell not for functional purposes but generating a competition with the functional partner , possibly participating to the regulation of the protein interactions in the cell ., Taking into account these possibilities , this set of highly potential partners becomes interesting for further studies ., For instance , these interactions would deserve to be experimentally tested to see how strongly they interact , and whether they form a structurally well-defined complex ., Also , for a given protein and a set of highly potential partners , one could ask whether general structural ( geometrical or physico-chemical ) features of the interface exist and in the positive case , classify these interfaces ., These further studies could contribute to give important insights into protein partnership discrimination ., For each protein , we defined a signature representing the strength of interaction of with all other proteins ., As mentioned above , signatures found for all s in the Mintseris dataset demonstrate the existence of strong interactions with some proteins , but also the absence of interactions with other proteins , and so on ., The spectrum of strengths of interactions suggests the notion of PPI to be revisited so to include the larger panel of potential complex formations between a protein and its potential partners ., Several questions could then be asked on proteins presenting similar signatures 44 , but they go beyond the aim of this work ., We have shown that evolutionary information can also be used to restrict the conformational space of the docking exploration without an important loss in sensitivity ., This result is very important in view of reducing the computational cost of highly time demanding docking calculations ( all atom description and precise energy functions ) and the perspective of enlarging the dataset size for future CC-D calculations ., To conclude , we are the first to perform a CC-D of a pool of proteins covering a large spectrum of functions and interaction modes , performing it on unbound structures and providing energy values ( even though simplified ) taking into account electrostatic forces ., Our approach is the first combining evolutionary information with CC-D simulations ., The evaluation of the performance of these two contributions to the problem of partner identification , suggests that there is still room for improvement in the solution ., In particular , we have shown that a precise identification of protein binding sites allows for very satisfactory predictions ., Data coming from the CC-D calculations and the evolutionary analysis are provided and they will help the community to evaluate further CC-D studies and methodological developments ., In particular , the decoy set constitutes a unique dataset of “negative” partners ., For them , we provide about 2000 conformations and an associated coarse-grain energy score ., It might be extremely useful to suitably parametrize docking scoring functions , more refined than our coarse-grain scoring function , to discriminate partners ., In the context of this study , a subset of these decoy structures filtered by our coarse-grain scoring function could be re-scored for a better partnership evaluation by using a more refined score function better discriminating the interaction signals ., The Docking Benchmark 2 . 0 25 is constituted by 168 proteins belonging to 84 known complexes ., We used the unbound conformations of the proteins with the exception of 12 antibodies for which the unbound structure is unavailable ., For those , the bound structure is used instead ., Any reference to the proteins uses either their name or the Protein Data Bank ( PDB ) code 45 of the experimental complex they belong to with the or extension denoting a receptor or a ligand protein respectively ., For example , 1AY7_r and 1AY7_l refer to barnase ( receptor ) and barstar ( ligand ) in the barnase-barstar complex 1AY7 ., The coordinates for the bound and unbound structures of both receptor and ligand proteins are available in the PDB and can be found at http://zlab . bu . edu/zdock/benchmark . shtml ., Molecular docking is performed with the MAXDo ( Molecular Association via Cross Docking ) algorithm , developed for complete cross-docking ( CC-D ) studies 12 ., Since CC-D involves a much larger number of calculations than simple docking , we chose a rigid-body docking approach using a reduced protein model in order to make rapid conformational searches ., Surface residues are residues with at least of accessible surface ., Accessibility is calculated with NACCESS 2 . 1 . 1 53 with a probe size of =\u200a1 . 4 Å ., Interface residues are residues with a change of at least decrease in accessible surface area compared to the unbound protein ., In order to improve the quality of the predictions of protein interaction partners , in our earlier study we developed a normalized interaction index ( NII ) that takes into account whether a protein-protein interface involves amino acids belonging to a known interaction site 12 ., This information can potentially be obtained using predictive tools ( see below ) , but here we use the experimentally determined interfaces of the 84 binary complexes in the Docking Benchmark ., We however recall that all our docking trials involve unbound protein conformations ., For each protein partner in a given complex , we determine which fraction of the docked interface residues ( abbreviated as ) are found in the experimental interface for ( ) and ( ) ., Thus defining an overall fraction for the complex as ., It is important to notice that the formula can be computed from either experimental interfaces ( as defined above ) or predicted interfaces ( where prediction could be realized , for instance , with evolutionary information; see paragraph below ) ., The notion of “FIR” proposes a new concept for docking evaluation that can be used as an alternative to the usual docking metrics 54 originally designed to evaluate the accuracy of pairwise protein docking models ., While the measure denotes the coverage of the experimental interface , that is the sensitivity of the predicted interface , the FIR denotes the PPV of the predicted interface ., Also , for the measure , contacts are defined with respect to a 5 Å cutoff on the RMSD of heavy atoms , while for FIR , contacts are defined from a change of solvent accessibility ., For every protein pair , we calculate an energy-weighted optimal interaction index ( ) defined in Eq ., ( 1 ) ., To allow comparison among different partners we defined a normalized index by taking into account all of the four lines/columns that feature either or in the matrix as follows: ( 3 ) where is a symmetrized version of the interaction index and it is defined as: ( 4 ) where are the 168 proteins of our dataset ., values vary between and ., Values close to zero imply that two proteins cannot form an interface involving a significant fraction of the experimentally identified residues , or that interfaces involving these residues have poor interaction energies ., Values close to one indicate predicted interfaces with good energies and composed of experimentally identified residues ., For each protein , we define as predicted partner of , the protein that leads to ., We consider as true positives ( ) all the predicted pairs that belong to the Docking Benchmark 2 . 0 and as true negatives ( ) all the pairs that are correctly predicted as non interacting ., We define a False Positive Rate ( ) and the True Positive Rate ( ) to be and , where is the set of False Positives ( partners incorrectly predicted as interacting ) and is the set of False Negatives ( partners incorrectly predicted as non interacting ) ., The computation of and for various thresholds enables the Receiver Operating Characteristics ( ROC ) curve to be drawn ., The performance of the prediction is given by the resulting AUC ( Area Under the Curve ) value ., Values of and correspond to random and perfect predictions respectively ., AUC calculations were performed with the R package 55 ., Also , given a threshold on the NII values , we use five standard measures of performance: sensitivity , specificity , precision or positive predictive value , balanced -score and Matthews correlation coefficient where ., To predict protein partners without using any experimental information , we define a new FIR measure by combining docking and evolutionary information ., From FIR values , NII matrices are computed as above ., The interaction rank of a protein pair is defined to be the best rank of the pair among all the pairs that have either or as receptor ., This means that given a matrix , we look at the rank of the pair with respect to the values , that is the line indexed by , and at the rank of the pair with respect to the values , that is the line indexed by ., The best rank computed for each line is retained for the pair ., To restraint the conformational space of the docking algorithm , we combine JET interface predictions with MAXDo , in such a way that only surface regions containing residues predicted by JET will be analyzed by MAXDo ., To do so , for each docked orientation , we computed the center of mass of the ligand and defined the axis linking it to the center of mass of the receptor ., We remind that the position of the receptor is fixed ., Along this axis , we define an imaginary tube of radius r\u200a=\u200a2 . 9 Å ., For each ligand orientation , we check whether the interface of the resulting ligand-receptor complex involves residues predicted by JET or not ( Figure S53 in Text S1 ) ., Each residue is approximated to a point whose coordinates represent the average of the atoms coordinates ., The distance of this point from the axis of the tube , allows to establish whether the residue falls inside the tube or not , and therefore , whether the ligand orientation should be retained or not ., Strictly speaking , one should also use the scalar product between the vectors going from the receptor center of mass to the residue and to the ligand center of mass ( this product decides whether the residue lies on the side of the ligand-receptor interface ) ., We ask for just one single residue in the orientation interface to be within the tube to retain this latter ., CC-D of the Mintseris Enzyme-Inhibitors dataset was performed with HEX v6 . 3 using the shape complementarity based-only score 42 ., Docked conformations were clustered using a 3 Å cutoff and the best-scored conformations of the 500 first clusters were retained for the analysis ., A protocol similar to that described for MAXDo was applied to evaluate partner prediction based on HEX results ,, ( i ) by assuming knowledge of the experimental interfaces and, ( ii ) by crossing docking scores with evolutionary information ., All 500 conformations were considered for residue scoring based on docking and for protein interaction index calculation ., Parameter values are reported in Table S7 in Text S1 ., Given a protein , we searched in the Mintseris dataset for those proteins that have a homolog coming from the same species as ., Namely , for each , we searched with Blast ( E-value threshold at , alignment coverage ) for the set of sequences that are at least , or or identical to the original sequence ., This provides a set of species that we say to be representing ., We then checked that the species of is included in the set of the species representing ., Notice that the protocol does not necessarily provide the same answer when it is applied to the protein pairs or due to the non-symmetrical Blast result ., We release the first large decoy database comprising not only structures of true complexes but also structures of non-functional complexes potentially forming in the cell ., For the 28224 possible protein pairs ( involving the 168 proteins ) of the Mintseris Benchmark 2 . 0 , we considered about 2000 best ligand orientations ( represented on and angles as described above ) for each receptor ., We provide the associated decoys together with the corresponding energy values ., A program to reconstruct the PDB structure of the conformation given and angles is also provided ., For each protein in the Mintseris dataset , we also furnish the evolutionary analysis for the detection of the binding sites ., The download site is http://www . lgm . upmc . fr/CCDMintseris/ | Introduction, Results, Discussion, Methods | Large-scale analyses of protein-protein interactions based on coarse-grain molecular docking simulations and binding site predictions resulting from evolutionary sequence analysis , are possible and realizable on hundreds of proteins with variate structures and interfaces ., We demonstrated this on the 168 proteins of the Mintseris Benchmark 2 . 0 ., On the one hand , we evaluated the quality of the interaction signal and the contribution of docking information compared to evolutionary information showing that the combination of the two improves partner identification ., On the other hand , since protein interactions usually occur in crowded environments with several competing partners , we realized a thorough analysis of the interactions of proteins with true partners but also with non-partners to evaluate whether proteins in the environment , competing with the true partner , affect its identification ., We found three populations of proteins: strongly competing , never competing , and interacting with different levels of strength ., Populations and levels of strength are numerically characterized and provide a signature for the behavior of a protein in the crowded environment ., We showed that partner identification , to some extent , does not depend on the competing partners present in the environment , that certain biochemical classes of proteins are intrinsically easier to analyze than others , and that small proteins are not more promiscuous than large ones ., Our approach brings to light that the knowledge of the binding site can be used to reduce the high computational cost of docking simulations with no consequence in the quality of the results , demonstrating the possibility to apply coarse-grain docking to datasets made of thousands of proteins ., Comparison with all available large-scale analyses aimed to partner predictions is realized ., We release the complete decoys set issued by coarse-grain docking simulations of both true and false interacting partners , and their evolutionary sequence analysis leading to binding site predictions ., Download site: http://www . lgm . upmc . fr/CCDMintseris/ | Protein-protein interactions ( PPI ) are at the heart of the molecular processes governing life and constitute an increasingly important target for drug design ., Given their importance , it is vital to determine which protein interactions have functional relevance and to characterize the protein competition inherent to crowded environments , as the cytoplasm or the cellular organelles ., We show that combining coarse-grain molecular cross-docking simulations and binding site predictions based on evolutionary sequence analysis is a viable route to identify true interacting partners for hundreds of proteins with a variate set of protein structures and interfaces ., Also , we realize a large-scale analysis of protein binding promiscuity and provide a numerical characterization of partner competition and level of interaction strength for about 28000 false-partner interactions ., Finally , we demonstrate that binding site prediction is useful to discriminate native partners , but also to scale up the approach to thousands of protein interactions ., This study is based on the large computational effort made by thousands of internautes helping World Community Grid over a period of 7 months ., The complete dataset issued by the computation and the analysis is released to the scientific community . | null | null |
journal.ppat.1006942 | 2,018 | miR-126-5p by direct targeting of JNK-interacting protein-2 (JIP-2) plays a key role in Theileria-infected macrophage virulence | Theileria annulata is an apicomplexan parasite causing a widespread disease called tropical theileriosis that is endemic to North Africa , the Middle East , vast parts of India and China 1 ., The parasite can infect bovine B cells , but in the natural environment , predominantly infects macrophages ., Theileria-infected leukocytes are transformed into tumour-like leukemias that display uncontrolled proliferation and increased ability to disseminate and invade organs and tissues 2 , 3 ., As the transformed state of Theileria-infected leukocytes can be reversed by drug ( buparvaquone ) -induced parasite death , molecular events related to Theileria-induced host cell transformation have been proposed to have an epigenetic basis 4 ., Since attenuated vaccine lines used to fight tropical theileriosis are derived by long-term culture of virulent infected macrophages , and since many virulence traits can be restored by TGF-β2–stimulation of attenuated macrophages 5 , loss of Theileria-infected macrophage virulence such as activation of the Activator Protein 1 ( AP-1 ) transcription factor 6 , 7 may also have an epigenetic element 8 , 9 ., Micro-ribonucleic acids ( miRNAs ) are small ( 17–25 bases long ) single-stranded , non-coding RNAs 10 , 11 that modulate diverse biological processes by normally binding to the 3′-untranslated region ( 3’-UTR ) of target mRNAs , thus altering the post-transcriptional regulation of numerous genes 12–14 ., However , miRNA can also bind to 5-UTRs , introns and coding sequence of mRNA 15–17 ., Post-transcriptional control of gene expression by miRNAs is increasingly recognized as a central part of host/pathogen interactions 18 , 19 ., The role of miRNAs in bacterial 20 , 21 , viral 22 and protozoan 23 infections is now well established ., A role for miR-155 in the virulence of T . annulata-infected leukocytes occurs via its suppression of De-Etiolated Homolog 1 ( DET-1 ) expression that diminishes c-Jun ubiquitination 9 ., In order to obtain a global view of all bovine miRNAs expressed in different types of T . annulata-infected leukocytes and how they might contribute to parasite-induced leukocyte transformation and tumour dissemination we determined the miRNomes of infected macrophages ( both virulent and attenuated ) and infected B cells ( TBL20 and TBL3 ) versus non-infected B cells ( BL20 and BL3 ) ., Defining the miRNomes of different types of T . annulata-transformed leukocytes allowed us to observe that infection alters the expression of many known miRNAs ., However , to identify miRNAs implicated in Theileria-transformed leukocyte dissemination , as opposed to immortalisation , we queried our datasets only for miRNAs whose expression is activated by infection , but dampened upon loss of transformed macrophage dissemination ., Here , we characterise the role of miR-126-5p in T . annulata-induced leukocyte transformation and attenuation of infected macrophage dissemination ., The mixed lineage kinase dual leucine zipper kinase-1 ( DLK1 ) is an established target of miR-126-5p 24 ., DLK1 can selectively regulate the JNK-based stress response pathway via its interaction with the scaffolding protein JIP to form a specialized JNK signalling complex 25 ., JIP-1 and JIP-2 bind selectively to JNK , but not to other related MAP kinases including p38 26 , 27 and their over-expression causes cytoplasmic retention of JNK; thereby preventing its nuclear translocation and its ability to phosphorylate specific nuclear substrates such as c-Jun 26 ., Moreover , under basal conditions DLK1 is bound to JIP , but upon stimulation , DLK1 dissociates from JIP resulting in JNK translocation to the nucleus 28 ., Within the nucleus Jun and Fos family members form homo- and hetero-dimers as part of the AP-1 transcription factor that binds to specific DNA sequences to drive target gene expression; a notable example being mmp9 ., The matrix metallo-proteinase MMP9 is associated with metastasis 29 and AP-1 is implicated in various cellular processes such as apoptosis 30 , growth control 31 and cellular transformation 32 ., Upon activation , JNK translocates to the nucleus to phosphorylate c-Jun that is a key transcription factor in the virulence-associated hyper-dissemination phenotype of Theileria-transformed leukocytes 2 , 7 ., Importantly , upon attenuation of Theileria-infected macrophage dissemination JNK activity decreases resulting in reduced c-Jun phosphorylation and decreased AP-1-driven transcription of mmp9 33 ., We now demonstrate that T . annulata infection of B cells and macrophages leads to the up-regulation of miR-126-5p that ablates JIP-2 expression liberating cytosolic JNK1 to translocate to the nucleus and phosphorylate c-Jun ., Conversely , in attenuated macrophages miR-126-5p levels drop , JIP-2 complexes reform retaining JNK in the cytosol leading to reduced nuclear c-Jun phosphorylation , dampened AP-1-driven transcription of mmp9 and reduced traversal of matrigel ., Thus , high miR-126-5p levels contribute to Theileria-transformed leukocyte dissemination and reduced miR-126-5p levels contribute to attenuation of the virulent hyper-dissemination phenotype ., To identify miRNAs whose expression is altered upon infection by T . annulata we determined the miRNomes of both infected versus non-infected B cells and virulent versus attenuated macrophages ., The comparison between T . annulata-infected TBL20 B lymphocytes and their uninfected counterparts revealed potential involvement of many miRNAs in the transformation of the host cells , as reflected by the changes in their expression levels ( Fig 1A ) ., We analyzed the differential expression ( DE ) of our miRNA sequencing data using DESeq2 34 ., 115 miRNAs are differentially expressed in TBL20 as compared to BL20 with a cutoff of adjusted p value < 0 . 05 ., In order to increase the confidence level and limit our analysis to the most significantly DE miRNAs we used a second pipeline , baySeq 35 ., Both DESeq2 and baySeq are highly specific and sensitive tools for the detection of differential expression 36 ., We consider a miRNA differentially expressed ( DE ) following two criteria:, a ) fold change ( FC ) greater than 2 and, b ) adjusted p value less than 0 . 05 ( DESeq2 ) and FDR less than 0 . 1 ( baySeq ) ., Finally , we compared the lists of up- and down-regulated miRNAs from both DESeq2 and baySeq and retained only miRNAs identified as DE in both pipelines ( S1 Table ) ., In addition to TBL20 , we characterize the DE miRNAs in a different B cell line: TBL3 ., Similarly , we identified the DE miRNAs in infected TBL3 as compared to their uninfected counterparts , BL3 , using the same pipelines and criteria ( S2 Table ) ., The comparison between the lists of DE miRNAs in TBL20 and TBL3 shows that there are 19 common differentially expressed miRNAs: 9 up- and 10 down-regulated ., We followed the expression of these 19 DE miRNAs in virulent compared to attenuated macrophages ( Fig 1B ) ., This identified miR-126-5p as a miRNA upregulated after T . annulata infection of B lymphocytes in 2 independent cell lines ( TBL20 and TBL3 ) and down-regulated in attenuated macrophages that have lost their hyper-disseminating phenotype ., As expected for transformed leukocytes the biological functions of the DE miRNAs are annotated as being associated with “oncogenesis” , with the exception of miR-6526 and miR-30f that are not well characterized ., The reported functions of the DE miRNAs are therefore consistent with the cancer-like phenotype of T . annulata-infected leukocytes ., To confirm the sequencing results , we randomly selected 10 DE miRNAs , 5 up- and 5 down-regulated and verified their expression qRT-PCR ( Fig 1C , left ) ., All tested miRNAs , including miR-126-5p , confirmed the miRNA sequencing data for DE ., Unlike T . parva , the causative agent of East Coast Fever that results from infection and transformation of T and B cells , T . annulata-transformed macrophages lose their virulent hyper-disseminating phenotype following long-term culture , and attenuated macrophages with diminished dissemination are used as live vaccines against tropical theileriosis ., For these reasons , we used next generation sequencing ( NGS ) to profile the miRNomes of non-infected B cells , T . annulata-infected B cells and virulent versus attenuated infected macrophages ., The different miRNomes allowed us to compare the miRNA expression of B cells before and following infection and concomitant with loss of T . annulata-infected macrophage virulence ., To focus on miRNAs of particular relevance to parasite-induced leukocyte tumour virulence we asked that expression of a given miRNA be upregulated by infection yet downregulated in attenuated macrophages that have lost their hyper-disseminating phenotype ., These criteria identified miR-126-5p as a prime candidate and led to our characterization of its contribution to the transformed phenotype of T . annulata-infected B cells and macrophages ., We demonstrated that JIP-2 is a novel miR-126-5p target gene and that infection by increasing miR-126-5p levels suppresses JIP-2 expression in virulent macrophages ., Loss of JIP-2 released cytosolic JNK1 to translocate to the nucleus and phosphorylate c-Jun , contributing to constitutive AP-1-driven MMP production that is characteristic of Theileria-induced leukocyte dissemination ., By contrast , in attenuated macrophages , where miR-126-5p expression is reduced , augmented JIP-2 retains JNK1 in the cytosol leading to decreased nuclear c-Jun phosphorylation , ablated MMP9 production and dampened traversal of matrigel ., Thus , miR-126-5p-provoked reduction in JIP-2 levels activates JNK1>AP-1 signalling and provides an epigenetic explanation for both T . annulata-induced leukocyte transformation , and for the attenuated phenotype of live vaccines against tropical theileriosis ., By demonstrating that infection-induced miR-126-5p expression ablates JIP-2 and diminishes the cytosolic localisation of JNK1 we provide a mechanism that contributes to constitutive c-Jun phosphorylation , increased MMP9 production and a greater capacity of Theileria-transformed leukocytes to disseminate ( Fig 7 ) ., miR-126 is located within the 7th intron of the EGFL7 gene 37–39 and EGFL7 is equivalently expressed in virulent and attenuated macrophages ( S1 Fig ) ., In T . annulata-infected macrophages miR-126-5p levels therefore do not depend on the degree of EGFL7 expression 48 , nor on the amount of precursor miR-126 , rather infection impacts on the capacity of AGO2 to load miR-126-5p , where it’s protected from degradation , while miR-126-3p is not loaded and is consequently degraded ., In virulent macrophages Grb2 recruits PTP1B to de-phosphorylate AGO2 that facilitates uptake of miR-126-5p , whereas in attenuated macrophages the amount of PTP1B associated with AGO2 diminishes with a concomitant increase in AGO2 phosphorylation and decrease in bound miR-126-5p ( Fig 7 ) ., Inflammation stemming from T . annulata infection likely explains induction of EGFL7 and pre-miR-126 expression , but why miR-126-5p , rather than miR-126-3p , is loaded onto AGO2 is unknown and will animate future studies ., Finally , given that miR-126-5p is deregulated in many cancers; reagents that manipulate miR-126-5p levels could be discussed as tools for cancer therapy ., Cells used in this study are T . annulata-infected Ode macrophages 49 , where virulent macrophages used correspond to passage 62 and attenuated macrophages to passage 364 ., The different B cell lines used were non-infected immortalized B sarcoma lines ( BL3 and BL20 ) and T . annulata-infected BL3 ( TBL3 ) and BL20 ( TBL20 ) cells , and all have been previously described 50–53 ., All cells were incubated at 37°C with 5% CO2 in Roswell Park Memorial Institute medium ( RPMI ) supplemented with 10% Fetal Bovine Serum ( FBS ) , 2mM L-Glutamine , 100 U penicillin , 0 . 1mg/ml streptomycin , and 4- ( 2-hydroxyethyl ) -1-piperazineethanesulfonic acid ( HEPES ) and 5% 2-mercapthoethanol for BL20 and TBL20 ., Total RNA of Theileria-infected leukocytes was isolated using the miRNA isolation kit ( #AM1560 , Thermo fisher scientific , Villebon , France ) according to the manufacturer’s instructions ., Total RNA designated for miRNA experiments was extracted using the mirVana miRNA isolation kit ( Thermo Fisher ) using the manufacturer’s instructions ., The quantity and quality checked by Qubit and Bioanalyzer 2100 , respectively ., The miRNA libraries were prepared using the Illumina Truseq Small RNA Sample Preparation kit ( RS-200-0012 ) following the manufacturer’s instructions ., Briefly , 1 following the manufactupter ligated at the 3’ and 5’ ends , reverse transcribed , barcoded then amplified with 11 cycles of PCR amplifications ., Then the cDNA was run on a 6% TBE PAGE gel ( Novex , Thermo Fisher ) ., After staining with SYBR Green the gel is visualized on a UV transluminator ( Doc-It imaging system , UVP ) and the cDNA constructs of a size between 145–160 bp were cut out and eluted from the gel , concentrated and the libraries validated , quantified and finally pooled and sequenced on a Hiseq 2000 and Hiseq 4000 ., Raw miRNAseq reads are quality checked by fastqc 54 ., mirTools2 . 0 55 pipeline is used for trimming ( “Adaptor_trim . pl” script ) and downstream analysis of known miRNAs ., Sequencing reads are aligned to the bosTau7 genome using SOAP 56 ., Annotations are added from miRBase 21 57 and Rfam 58 databases ., The differential expression of each known miRNAs from their absolute read counts are analysed by DESeq2 34 ., Differential expression is tested based on the negative Binomial distribution and miRNAs with adjusted p-value < 0 . 05 considered as statistically significant ., Bovine genomic DNA of mapk8 ( jip2 ) was purified from Theileria-infected macrophages and used as a template for PCR ., For construction of the jip2 , the 3′-UTRs of bovine JIP2 was amplified using the following primers: Forward: GGCCCTCGAGAGCAGAAAGTTTATTGAGGTGCT Reverse: GGCCGCGGCCGCTGGGGTCGGAACTGGGAG The PCR-amplified fragments were digested by XhoI and NotI and inserted in the psiCHECK-2 vector ., Cells were transiently transfected with 2 μg of firefly/renilla luciferase reporter plasmid ., Protein extracts were prepared using the Passive Lysis Buffer provided in the Dual-Luciferase Assay ( Promega ) ., Equal amounts of protein extracts were plated into a 96-well plate ., Firefly luciferase activity was measured for 12 seconds using the LB 960 luminometer ( Berthold Technologies , Thoiry , France ) ., To assess the internal standard activity , Stop and Glo reagent was added ( Promega ) , and the peak of the renilla luciferase activity was then measured ., Normalized relative luciferase units ( RLU ) were then calculated as firefly luciferase units of protein extracts of treated or untreated cells divided by renilla luciferase units of protein extracts of untreated cells ., Data represent the mean ± SEM of three independent experiments , each performed in duplicate ., Cells were transfected by electroporation using the Nucleofector system ( Lonza , Basel , Switzerland ) ., 2 . 5x105 cells were suspended in 100μl of nucleofector solution mix with 2μg of plasmids and subjected to electroporation using the cell line—specific programme: T-O17 ., After transfection , cells were suspended in fresh complete medium and incubated at 37°C , 5% CO2 for 24 h and RNA extracted after 48 h post transfections ., Measurements of luciferase and β-galactosidase activities were performed using the Dual Light Assay system ( Thermo Fisher scientific ) and luminometer Centro LB 960 ( Berthold ) according to the manufacturer’s instructions ., RNA extracted with the mirVana kit was used ., cDNA was synthesized using the miScript II RT kit ( Qiagen ) following the manufacturer’s instructions ., Briefly , a 20 μl was set for each biological replicate of each tested miRNA ., The RT contained 8 μl MMIX ( 4 μl of 5x miScript HiFlex Buffer , 2 μl of 10x miScript Nucleics Mix and 2 μl miScript Reverse Transcriptase Mix ) , 500 ng of total RNA in RNase-free water ., The RT was performed at 37°C for 60 min and 95°C for 5 min ., The qPCR was performed in a 96-well plate as a 25 μl volume containing 2μl of RT product , 12 . 5 μl of 2x QuantiTect SYBR Green PCR Master Mix , 2 . 5 μl of 10x miScript Precursor Assay and RNase-free water ., The qPCR thermal cycle was set to 95°C for 15 min and 40 cycles of 94°C for 15 secs , 55°C for 30 secs and 70°C for 30 sec ., Data were analysed using the 2−ΔΔCT , or the relative expression method by normalization to HRPT ( ENSBTAT00000019547 . 4 ) as a reference gene ., RNA extracted with the mirVana kit was used ., cDNA was synthesized using the TaqMan microRNA RT kit ( Thermo Fisher scientific ) following the manufacturer’s instructions ., Briefly , a 15 μl was set for each biological replicate of each tested miRNA ., The RT contained 7 μl MMIX ( 100 mM dNTPs , 50 U/μl MultiScribe reverse transcriptase , 10X RT buffer , 20 U/μl RNase inhibitor ) , 10 ng of total RNA in 5 μl and 3 μl of 5X miRNA-specific RT primers ., The RT was performed at 16°C for 30 min , 42°C for 30 min and 85°C for 5 min ., The qPCR was performed in a 384well plate as a 10 μl volume containing 1 . 33 μl of RT product , 5 μl TaqMan 2X universal PCR MMIX , 1 μl of miRNA-specific 20X TaqMan MicroRNA assay ., The qPCR thermal cycle was set to 95°C for 10 min and 40 cycles of 95°C for 15 secs and 60°C for 60 sec ., Data was analysed using the 2−ΔΔCT , or the relative expression method by normalization to U6b as a reference gene for miRNA ., bta-miRNA primers were purchased from Thermo Fisher Scientific ., Total RNA of Theileria-infected leukocytes was isolated using the RNeasy mini kit ( Qiagen ) , according to the manufacturer’s instructions ., The quality and quantity of RNA were measured by Bioanalyzer 2100 and Qubit , respectively ., For reverse transcription , 1μg isolated RNA was diluted in water to a final volume of 12 μl , warmed at 65°C for 10 min , then incubated on ice for 2 min ., Afterwards , 8 μl of reaction solution ( 0 . 5 μl random hexamer , 4 μl 5x RT buffer , 1 . 5 μl 10mM dNTP , 1 μl 200U/μl RT-MMLV ( Promega , Charbonnières-les-Bains , France ) and 1 μl 40U/μl RNase inhibitor ( Promega ) was added to get a final reaction volume of 20 μl and incubated at 37°C for 2 h ., The resultant cDNA was stored at -20°C ., mRNA expression levels were estimated by qPCR on Light Cycler 480 ( Roche , Meylan , France ) using SYBR Green detection ( Thermo Fisher Scientific ) ., The detection of a single product was verified by dissociation curve analysis and relative quantities of mRNA calculated using the method described by 59 ., gapdh was used as reference gene to normalize for mRNA levels ., The specificity of PCR amplification was confirmed by melting curve analysis ., Sequence primers used are as follows: gapdh: FO 5’-AGGACAAAGCTCAGGGACAC-3’ , Rev 5’- CCCCAGGTCTACATGTTCCA-3’ mmp9: FO 5’-CCCATTAGCACGCACGACAT-3’ , Rev 5’- TCACGTAGCCCACATAGTCCA-3’ dlk1: FO 5’- ATGGGCATCGTCTTCCTCAA -3’ , Rev 5’- CAGGATGGTGAAGCAGATGG -3’ jip2: FO 5’- TCTTCCCTGCCTTCTATGCC -3’ , Rev 5’- CAGGTGGACGGTCAGTTT -3’ For total cell extraction , cells were harvested and extracted by lysis buffer ( 20mM Hepes , Nonidet P40 ( NP40 ) 1% , 0 . 1% SDS , 150mM NaCl , 2mM EDTA , phosphatase inhibitor cocktail tablet ( PhosSTOP , Roche ) and protease inhibitor cocktail tablet ( Complete mini EDTA free , Roche ) ) ., For cytoplasmique extraction , cells were harvested and extracted by lysis buffer ( HEPES 10 mM pH 7 . 9 , KCl 10 mM , EDTA 0 . 1 mM , NP-40 0 . 3% , protease inhibitors 1x , protease and phosphatase inhibitor cocktail ) ., For nuclear extraction , cell pellets were lysed and extracted by lysis buffer ( HEPES 20 mM pH 7 . 9 , NaCl 0 . 4 M , EDTA 1mM , Glycerol 25% and protease Inhibitors 1x ., Protein concentration was determined by the Bradford protein assay 60 ., Cell lysates were subjected to Western blot analysis using conventional SDS/PAGE and protein transfer to nitrocellulose filters ( Protran , Whatman ) ., The membrane was blocked by 5% non-fat milk-TBST ( for anti-DLK , anti-JIP-2 , anti-c-JUN , anti-JNK ) , or 3% non-fat milk-PBST ( for anti-actin antibody ) for 2 h at room temperature ( RT ) ., Antibodies used in immunoblotting were as follows: goat polyclonal antibody anti-JIP-2 ( Santa Cruz Biotechnologies , Heidelberg , Germany # sc-19740 ) , rabbit polyclonal antibody anti-JIP-2 ( Abcam # ab-154090 ) , rabbit polyclonal antibody anti-DLK ( Santa Cruz Biotechnologies # sc-25437 ) , rabbit polyclonal antibody anti-JNK ( Santa Cruz Biotechnologies # sc-571 ) , goat polyclonal anti-PTP1B ( Santa Cruz Biotechnologies # sc-1718 ) , mouse monoclonal antibody anti-AGO2 ( Abcam # ab-57113 ) , rabbit monoclonal anti-AGO2 ( Cell signalling # 2987 ) , mouse anti-phospho tyrosine antibody ( Transduction laboratories # P1120 ) , goat anti-GST antibody ( GE Healthcare # 27-4577-01 ) and goat polyclonal antibody anti-actin ( Santa Cruz Biotechnologies I-19 ) ., After washing , proteins were visualized with ECL western blotting detection reagents ( Thermo Scientific ) ., The β-actin level was used as a loading control throughout all experiments ., Co-immunoprecipitations and GST-Grb2 pull down assay were conducted with protein extracts of Theileria-infected macrophages ., JIP-2 , AGO2 , GST-Grb2 and PTP1B precipitates were transferred to western blots and probed respectively with a rabbit polyclonal anti-DLK , mouse monoclonal anti-AGO2 , mouse anti-phospho tyrosine , rabbit anti-PTP1B ( Abcam #ab88481 ) and goat anti-PTP1B antibodies ., Normal IgG was used as a negative control , cell lysates from virulent and attenuated macrophages were treated with IgG and the whole cell lysate without IP was included as positive control ., 1×105 cells were centrifuged on glass slide with the Cellspin I ( Tharmac ) at 1500 rpm for 10 min and fixed by 4% paraformaldehyde for 10–15 min at room temperature ., Fixed cells were permeabilized by 0 . 2% Triton X-100 for 10 min and blocked with 1% BSA for 30 min ., These cells were incubated with primary antibodies against Ser-phospho-73 c-Jun ( 1/200 , Santa Cruz Biotechnologies #sc-7981 ) overnight , sequentially stained with secondary antibodies conjugated with Alexa 488 ( 1/1000 , Molecular Probes ) for 45 min at room temperature ., Stained cells were mounted in ProLong Diamond Antifade Mountant with DAPI ( Thermo Fisher Scientific ) ., Acquisitions were made by inverted microscopy ( Leica DMI6000s ) with metamorphous software ., Images were taken at x100 magnification ., The invasive capacity of Theileria-infected macrophages and B cells were assessed in vitro using matrigel migration chambers 7 ., Culture coat 96-well medium BME cell invasion assay was obtained from Culturex instructions ( 3482-096-K ) ., Fifty thousand cells were added to each well and after 24 h of incubation at 37°C , each well of the top chamber was washed once in buffer ., The top chamber was placed back on the receiver plate ., 100 μl of cell dissociation solution/Calcein AM were added to the bottom chamber of each well , incubated at 37°C for 1 h to fluorescently label cells and dissociate them from the membrane before reading at 485 nm excitation , 520 nm emission using the same parameters as the standard curve ., Data were analysed with the Student’s t-test ., All values are expressed as mean+/-SEM ., Values were considered to be significantly different when two-sided p values were < 0 . 05 ., The miRNA expression dataset has been assigned series record GSE97706 in the GEO repository ., The work was conducted under the approval number 15IBEC11 by the Institutional Biosafety and Ethics Committee ( IBEC ) in KAUST . | Introduction, Results, Discussion, Materials and methods | Theileria annulata is an apicomplexan parasite that infects and transforms bovine macrophages that disseminate throughout the animal causing a leukaemia-like disease called tropical theileriosis ., Using deep RNAseq of T . annulata-infected B cells and macrophages we identify a set of microRNAs induced by infection , whose expression diminishes upon loss of the hyper-disseminating phenotype of virulent transformed macrophages ., We describe how infection-induced upregulation of miR-126-5p ablates JIP-2 expression to release cytosolic JNK to translocate to the nucleus and trans-activate AP-1-driven transcription of mmp9 to promote tumour dissemination ., In non-disseminating attenuated macrophages miR-126-5p levels drop , JIP-2 levels increase , JNK1 is retained in the cytosol leading to decreased c-Jun phosphorylation and dampened AP-1-driven mmp9 transcription ., We show that variation in miR-126-5p levels depends on the tyrosine phosphorylation status of AGO2 that is regulated by Grb2-recruitment of PTP1B ., In attenuated macrophages Grb2 levels drop resulting in less PTP1B recruitment , greater AGO2 phosphorylation , less miR-126-5p associated with AGO2 and a consequent rise in JIP-2 levels ., Changes in miR-126-5p levels therefore , underpin both the virulent hyper-dissemination and the attenuated dissemination of T . annulata-infected macrophages . | Theileria annulata-infected bovine macrophages lose their hyper-disseminating virulent phenotype during long-term culture and are used as attenuated live vaccines to fight tropical theileriosis ., Deep microRNA sequencing revealed that infection of both B cells and macrophages alters the expression of a large number of host cell microRNAs ., We focused on miR-126-5p as its expression was induced by infection , but diminished in attenuated macrophages that had lost their disease causing disseminating phenotype ., We show that miR-126-5p in virulent macrophages directly targets and suppresses a cytosolic scaffold protein called JNK-Interacting Protein-2 ( JIP-2 ) , so liberating JNK1 to enter the nucleus and phosphorylate c-Jun ., This activates AP-1-driven transcription of mmp9 that promotes tumour dissemination ., In virulent macrophages , an adaptor protein called Grb2 recruits the tyrosine phosphatase PTP1B to AGO2 so decreasing AGO2 phosphorylation to increase miR-126-5p levels ., By contrast , in attenuated macrophages AGO2 tyrosine phosphorylation increases and miR-126-5p levels drop leading to a regain in JIP-2 expression that retains JNK1 in the cytosol ., This leads to decreased nuclear c-Jun phosphorylation and reduced mmp9 production ., Thus , variations in miR-126-5p levels underpin both virulent hyper-dissemination and attenuation of T . annulata-transfected macrophages . | blood cells, transfection, phosphorylation, medicine and health sciences, immune cells, chemical compounds, gene regulation, immunology, organic compounds, micrornas, tyrosine, immunoprecipitation, amino acids, molecular biology techniques, research and analysis methods, aromatic amino acids, white blood cells, animal cells, proteins, gene expression, chemistry, molecular biology, precipitation techniques, biochemistry, rna, antibody-producing cells, cell biology, nucleic acids, b cells, post-translational modification, organic chemistry, genetics, hydroxyl amino acids, biology and life sciences, cellular types, physical sciences, macrophages, non-coding rna | null |
journal.pntd.0006103 | 2,017 | The importance of thinking beyond the water-supply in cholera epidemics: A historical urban case-study | Cholera transmission during an outbreak is known to occur via both ‘short-’ ( for example , locally-mediated via food or household water ) 1–3 and ‘long-cycles’ ( environmentally-mediated via natural or manmade water and sanitation systems ) 4 , 5 ., Although long-cycle waterborne transmission is often considered the archetypical cholera transmission route 6 , 7 , there is a growing interest in understanding the importance of short-cycle pathways 8–10 ., Identifying the relative contributions of these different pathways has important public health implications for designing effective cholera interventions 11 , yet remains difficult to ascertain from epidemiological data alone 12–14 ., The proportion of cases infected via long-cycle relative to short-cycle transmission is probably highly context dependent , but the difficulty in distinguishing between the relative contributions of each route means that little data exists for any individual context 15 , 16 ., The resulting uncertainty hinders planning of appropriate interventions ., For example , spatially targeted interventions ( e . g . targeted hygiene/sanitation or reactive vaccination programs ) may be effective against short-cycle transmission 17 , but suboptimal against long-cycle transmission 18 ., To better characterize the relative importance of long-cycle vs short-cycle cholera transmission and mitigate parameter identifiability issues , high-resolution , high-quality epidemiological data can be augmented with data detailing the flow of drinking water in a specific setting ., We use detailed data from an 1853 cholera outbreak in Copenhagen , Denmark as a case example ., This outbreak has three key advantages from a modeling perspective: ( 1 ) this was likely an immunologically naive population as this was the first reported cholera outbreak in Copenhagen , ( 2 ) the outbreak was largely unmitigated by control measures as no effective treatments or interventions were implemented , and ( 3 ) historical datasets provide detailed information about the city’s hydraulic network ., The water supply of Copenhagen was composed of a network of hollowed wooden tree-trunks under low water-pressure , and thus vulnerable to outside contamination ( S1 Fig ) ., Additionally , there was no sewage system; rather , the street gutters functioned as light sewage drainage , with most human waste stored in unsealed cellars in each building and removed by night-men twice annually for use as fertilizer for food crops in nearby fields 19 ., The piped drinking water was reported to be contaminated by seepage from these cesspools 20 , thus elevating the risk of drinking contaminated water for downstream users during the cholera epidemic ., Most piped water was sourced from nearby ( <1 km ) semi-artificial lakes , while some public fountains were supplied from water sources >10 km upstream 21 ., Here , we use newly uncovered historical epidemiologic data from the 1853 epidemic along with modern statistical methods to investigate the relative contribution of short-cycle versus long-cycle transmission in a cholera outbreak ., Fitting a time-series meta-population model to data from the 1853 cholera outbreak in Copenhagen , we characterize the spatio-temporal transmission dynamics and assess the signal of long-cycle environmentally-mediated transmission in the progression of the epidemic ., Weekly cholera morbidity and mortality data for each city neighborhood and outlying communities were obtained from datasets compiled by contemporary physicians conducting active surveillance during the 1853 cholera epidemic in Copenhagen 22 ., A corps of physicians traveled door-to-door to residences , hospitals ( “sick-houses” ) , and “poor-houses” diagnosing and tabulating cases ., Hospitalized cases were assigned to the neighborhood of residence , unless they were already in hospital prior to diagnosis , in which case they were geolocated to the hospital’s address ., Six of the 13 neighborhoods were aggregated into two neighborhoods labeled “Combined upper” ( Nørre and Klædebo ) and “Combined lower” ( Frimands , Strand , Snarens and Vester ) due to low case counts and small geographic area ., We excluded 621 ( 9% ) cases that could not be geolocated , consisting of 122 ( 2% ) cases from docked ships , and 499 ( 7% ) from scattered outlying communities ., Cholera cases were defined as patients with rice-water diarrhea and evidence of severe dehydration 23 , making the historical diagnostic criteria stricter than the current WHO suspected cholera case definition 24 ., Population size for each neighborhood was interpolated between the 1850 and 1855 census assuming a linear growth model ., Population density was estimated for each neighborhood by georeferencing the neighborhoods using a Geographic Information System ( GIS ) and calculating area ., Hydraulic data was digitized from a contemporary map 21 showing the layout and direction of flow of all water-pipes supplying drinking water for the city ( Fig 1A ) ., We created a binary asymmetric transition matrix of hydraulic connectivity describing the flow of water between neighborhoods , and a binary symmetric matrix describing neighborhoods that share a border ( as a proxy for sewage runoff and human connectivity ) ( S1 Table and S2 Table ) ., All GIS work was done in QGIS 25 ., To model the number of infectious people at each day , we fit a discrete time susceptible-infected-recovered ( SIR ) model to imputed daily case data We used the basic model structure from Azman et al . 17 , with modifications as described below ., In order to simulate daily case data ( necessary because of the short generation interval for cholera ) , weekly cases were randomly reallocated to the seven days preceding the reported date ., This reallocation was repeated 10 times to give 10 possible realizations of the epidemic ., Each of the nine neighborhoods were considered as a discrete population ., The model-predicted number of infected people in each neighborhood at each time-step was Poisson distributed where the mean was a function of the fraction susceptible in that neighborhood and a sum of the internal and external forces of infection ., We constructed a series of nested models in which the infection rate of susceptible individuals living in neighborhood i , or the force of infection , λi , was the sum of an internal force of infection , βi , and an external force of infection from neighborhood j upon neighborhood i , αj , i ( Fig 2 ) , such that:, λi=βiIi+∑j≠iαj , iIj, Each model had differing assumptions about α and β ., Starting from the simplest base-case model to the most complex saturated model , we allowed: ( 1 ) a single β and single α for all neighborhoods such that βi = β and αj , i = α , ( 2 ) an individual βi for each neighborhood and a single α for all neighborhoods such that αj , i = α , and ( 3 ) an individual βi for each neighborhood and an asymmetric αj , i such that αj , i ≠ αi , j ( S1 Text ) ., In order to estimate the effect of the water supply on the epidemic , we used two methods ., First , we fit a linear regression model using the log of the median cross-neighborhood transmission coefficients ( αj , i ) from model 3 ( saturated model ) as the outcome and tested whether the hydraulic-transition and geographic-proximity variables were significant predictors of between-neighborhood transmission at a significance level of 0 . 05 ( S1 Text , S1 Table and S2 Table ) ., Second , we incorporated the hydraulic-connectivity and geographic-proximity matrices into model 2 , producing models 2b and 2c , respectively ., To assess the effect of hydraulic connectivity ( model 2b ) , we allowed αj , i to vary depending on whether the neighborhoods were connected via water pipes , such that:, αj , i={α0ifnowaterconnectionj→iα0+α1ifwaterconnectionj→i, To incorporate geographic proximity ( model 2c ) , we added an additional term if the neighborhoods shared a border ., The resulting αj , i can be described as:, αj , i={α0ifnosharedborderorwaterconnectionj→iα0+α1ifnosharedborderbutwaterconnectionj→iα0+α1+α2ifsharedborderandwaterconnectionj→i, Parameter estimation was done using Markov chain Monte Carlo ( MCMC ) methods in JAGS 4 . 2 . 0 26 in R 3 . 3 . 1 27 ., Each model was run once on each of the 10 possible realizations of the imputed daily epidemic data ., We used four MCMC chains with a burn-in of 40 , 000 iterations and sampled the subsequent 60 , 000 iterations ., Chain convergence was assessed with a potential scale reduction factor ( R^ ) cutoff of 1 . 05 and a visual assessment of the trace plots ., Posterior samples from all four chains and all 10 data realizations were pooled and summarized ( S1 Text ) ., Model preference was assessed using the average Watanabe-Akaike Information Criterion ( WAIC ) ; a difference of five in WAIC was considered significant , translating to a fold difference of e5/2 = 12 . 2 in terms of posterior probability on the model state space 28 ., WAIC was chosen due to recent developments in this field suggesting WAIC is better able to capture the tradeoffs between model complexity and fit in a Bayesian framework than the more commonly used DIC 29 , 30 ., We validated our model by simulating the outbreak in each neighborhood ., To do so , we selected a parameter set from the joint posterior distribution and then drew from the number of cases at the next time step from the appropriate Poisson distribution in each neighborhood ., Point-wise prediction intervals were constructed by taking the 2 . 5 and 97 . 5 percentiles at each time-step ., We then used the simulated data to refit the model parameters to test if the model could recapture the original parameter estimates ., To assess the heterogeneity of transmission efficiency by neighborhood , we calculated the internal reproductive number ( Rint ) , the outflowing reproductive number ( Rout ) , the inflowing reproductive number ( Rin ) , and the total reproductive number ( Rtot ) within each neighborhood using the force of infection coefficients ., The Rint , Rout , and Rtot can be interpreted as the number of cases a single infectious case produces within its own neighborhood , in all other neighborhoods , and in the whole city respectively , while the Rin represents the number of cases produced within one neighborhood by a single infectious case in the rest of the city ., The epidemic began on June 12 with two reported cases in the Nyboder neighborhood among individuals working on ( or in close contact with ) ships ., The epidemic soon spread to other neighborhoods ( Fig 3A and 3B , S1 Dataset ) ., The epidemic was declared over on October 1 , although four cases were reported subsequently in October ., A total of 7 , 173 cases were reported , of which 5 , 953 ( 83% ) were community acquired and 1 , 220 ( 17% ) were acquired in hospitals or poor-houses ., A total of 4 , 717 died , resulting in a case fatality ratio ( CFR ) of 66% ., The CFR among hospital/poor-house-acquired infections was 77% , as compared to 63% in community-acquired infections ., A total of 6 , 552 cases ( 91% ) occurred within the city walls and were included in this analysis ., The outbreak was spatially heterogeneous over the city , despite its small geographic size ., Attack rates within the city walls ranged from 2 . 1 to 9 . 6 per 100 , and the CFR ranged from 76% to 54% ( Table 1 , Fig 1C and 1D ) ., The neighborhood attack rates were not associated with the neighborhood’s population density as assessed by a linear regression ( p = 0 . 99 ) ., Model selection carried out using WAIC ( Table 2 ) indicated the best model allowed for asymmetric and heterogeneous transmission between neighborhoods ( model 3 ) ; this heterogeneity could not be explained by hydraulic and geographic connectivity ( model 2b and 2c ) ., Similarly , the linear regression on the median of the cross-neighborhood transmission coefficients ( αj , i ) ( Fig 4 ) from the fully saturated model ( model 3 ) provided no support that a hydraulic connection between neighborhoods was significantly associated with the force of transmission between neighborhoods , even after controlling for geographic proximity ( Table 3 ) ., The internal forces of transmission in each neighborhood were not strongly associated with the neighborhood’s population density ( p = 0 . 11 ) ., Using the fully saturated model ( model 3 ) , we simulated the outbreak in each neighborhood one day ahead , drawing a new parameter combination from the joint posterior distribution for each simulation ( Fig 5 ) ., The Rint ranged from 0 . 2 ( 0 . 0–0 . 5 ) in Østerto to 1 . 7 ( 1 . 5–1 . 8 ) in St . Annæ Vester ( Fig 6 ) ., One of the nine neighborhoods ( St . Annæ Vester ) , had Rint >1 , meaning it could maintain epidemics without infections from outside; in three other neighborhoods ( St . Annæ Øster , Christianshavn and Nyboder ) , Rint was above 1 but the credible interval spanned 1 . 0 , and in all remaining neighborhoods the Rint was below 1 . 0 ., To validate the model , we re-estimated the model parameters from the simulated data for each of the 10 epidemic realizations ., In seven of the 10 realizations , the 95% CI estimated from the simulated data overlapped 95% CI fitted from the original data for all 83 parameters ., In the remaining three realizations overlap occurred in 82 , 81 and 80 parameters respectively ., There is growing momentum towards re-thinking the dominance of long-cycle waterborne transmission in cholera outbreaks 31 ., Using a spatially explicit metapopulation model , we captured the essential spatial-temporal dynamics of a major urban cholera epidemic in Copenhagen in 1853 ., Our analysis indicates that although transmission occurred between the different sections of the city , the data do not support an association between the trajectory of cases across neighborhoods and the flow of water/sewage from neighborhood to neighborhood ., The lack of a signal of long-cycle transmission in the data , suggests the importance of short-cycle transmission ., The data do not match what would be expected for an epidemic primarily disseminated via long-cycle , waterborne transmission ., There are three pieces of evidence for this: ( 1 ) we found the attack rates to be highly heterogeneous within the city despite all neighborhoods sharing common sources for drinking water and extensive water-pipe connections between the neighborhoods; ( 2 ) neighborhoods , such as Øster , Kjøbmager and Combined Lower , that were downstream of highly affected neighborhoods , such as St . Annæ Vester and Nyboder , and ( 3 ) a model fit to the between-neighborhood transmission coefficients from the fully saturated model ( model, 3 ) did not show evidence that water-flow between neighborhoods was associated with the force of infection between neighborhoods ., The transmission heterogeneity seen in Copenhagen has been documented in other cholera outbreaks over a range of spatial scales 17 , 32 and was unlikely to be confounded by socioeconomic status ( SES ) ., We suspect this because SES did not have a spatial structure at the city level; the rich and poor lived on the same city blocks with the rich living facing the streets and the poor inside interior courtyards 33 ., Furthermore , population density was not associated with attack rate or the force of internal transmission ., In terms of differential reporting , the relative uniformity of the case fatality ratio ( Fig 1D ) implies that all neighborhoods were similarly likely to report cases ., This suggests that the transmission heterogeneity seen is a true phenomenon rather than an artifact of confounding ., The lack of support for long-cycle waterborne transmission in the Copenhagen epidemic has important public health implications for responding to present-day cholera outbreaks ., Our analysis indicates that interrupting transmission by interventions targeting the centralized drinking water system would likely have had little effect ., Despite the historical nature of our data , the Copenhagen outbreak can be a proxy for contemporary resource-constrained settings where water infrastructure is poor ., Our results corroborate other research , both in historical and contemporary settings , where short-cycle transmission was described as a critical element in certain epidemic settings 1 , 10 , 31 , 34 , 35 ., Taken together we propose public health practitioners need to more thoroughly investigate alternative transmission routes when investigating cholera outbreaks and planning interventions ., A stronger focus on interrupting short-cycle transmission , including targeted reactive vaccination programs and sanitation and hygiene related interventions , could significantly reduce an epidemic’s impact 36–38 ., There are several limitations to this analysis ., Although we found no support for long-cycle transmission driving transmission between neighborhoods , our analysis could not rule out this transmission pathway entirely ., Our results show some transmission did occur between neighborhoods , although it did not correlate with the piped water supply or neighborhood proximity ( as a proxy for sewage run-off ) ., Additionally , despite the high spatial-resolution of our data , long-cycle transmission within a single neighborhood is possible and would instead be captured by the within-neighborhood transmission coefficient ( βi ) ., It is probable that the epidemic resulted from multiple disparate transmission routes , a hypothesis consistent with recent models of the Haitian outbreak 34 , 39 ., Our analysis could not investigate any specific alternative pathway , although contemporary reports of human waste used as fertilizer for food crops highlights one possible alternative route , which has been observed in other outbreaks as well 40 ., Additionally , the available incidence data did not have the ideal temporal resolution for the analysis; it was aggregated into weekly time-steps , yet the generation time for cholera is suspected to be closer to 3–5 days 41 ., To address this limitation , we randomly redistributed the weekly cases over the preceding week , thereby propagating uncertainty in the relationship among weekly-aggregated cases in terms of who transmitted to whom ., In regard to hydraulic data , our measure of hydraulic connectivity was reduced to a binary value and does not fully capture the gradations of neighborhood water connections ., Although we argue the 1853 Copenhagen outbreak was not driven by long-cycle transmission , it may have played a larger role in outbreaks in other settings , at least in their initial stages 5 , 42–44 ., In Haiti , for example , there is evidence that the initial outbreak , along with the continuing transmission , entailed significant long-cycle transmission based on hydrological transport of the pathogen 4 , 35 , 45 , 46 ., The factors determining which mode of transmission will dominate new epidemics are not understood , but perhaps consists of a combination of cultural and environmental factors ., The ability to classify the dominant transmission mode early on in an epidemic will be key to enacting effective contagion control strategies ., Using a unique and highly detailed dataset , we have shown that that long-cycle , environmentally-mediated transmission is not a prerequisite for explosive , large-scale cholera outbreaks ., While an exact quantification of each pathway’s contribution remains difficult , our findings—taken together with previous research 10 , 17 , 31 , 35 , 36 , 38—suggest spatially-targeted cholera interventions , such as reactive vaccination and hygiene/sanitation programs , are important tools to combat epidemics with significant short-cycle transmission ., Moreover , programs targeting long-cycle waterborne transmission may not be effective in all outbreak settings . | Introduction, Methods, Results, Discussion | Planning interventions to respond to cholera epidemics requires an understanding of the major transmission routes ., Interrupting short-cycle ( household , foodborne ) transmission may require different approaches as compared long-cycle ( environmentally-mediated/waterborne ) transmission ., However , differentiating the relative contribution of short- and long-cycle routes has remained difficult , and most cholera outbreak control efforts focus on interrupting long-cycle transmission ., Here we use high-resolution epidemiological and municipal infrastructure data from a cholera outbreak in 1853 Copenhagen to explore the relative contribution of short- and long-cycle transmission routes during a major urban epidemic ., We fit a spatially explicit time-series meta-population model to 6 , 552 physician-reported cholera cases from Copenhagen in 1853 ., We estimated the contribution of long-cycle waterborne transmission between neighborhoods using historical municipal water infrastructure data , fitting the force of infection from hydraulic flow , then comparing model performance ., We found the epidemic was characterized by considerable transmission heterogeneity ., Some neighborhoods acted as localized transmission hotspots , while other neighborhoods were less affected or important in driving the epidemic ., We found little evidence to support long-cycle transmission between hydrologically-connected neighborhoods ., Collectively , these findings suggest short-cycle transmission was significant ., Spatially targeted cholera interventions , such as reactive vaccination or sanitation/hygiene campaigns in hotspot neighborhoods , would likely have been more effective in this epidemic than control measures aimed at interrupting long-cycle transmission , such as improving municipal water quality ., We recommend public health planners consider programs aimed at interrupting short-cycle transmission as essential tools in the cholera control arsenal . | John Snow’s seminal work on the London cholera epidemic and Broadway pump helped establish cholera as a quintessential waterborne ( long-cycle ) pathogen ., However , there is renewed interest in the role that short-cycle ( e . g . food-borne and household ) transmission plays in epidemic contexts ., The distinction between transmission pathways can be important , as they may demand different interventions ., However , disentangling these pathways requires high-quality epidemiological and contextual data that is rarely collected in outbreak situations ., Here we use detailed cholera incidence and municipal infrastructure data from a cholera outbreak in 1853 Copenhagen to estimate the relative contributions of short- and long-cycle transmission to the epidemic ., We find transmission between neighborhoods during the epidemic did not follow water pipe connections , suggesting little evidence of long-cycle transmission ., Instead , we suggest that short-cycle transmission was likely critical to the propagation of the outbreak ., Interventions targeting short-cycle transmission are important tools that merit further consideration by public health officials combating epidemic cholera . | medicine and health sciences, ecology and environmental sciences, water resources, surface water, engineering and technology, sewage, neighborhoods, tropical diseases, social sciences, simulation and modeling, vaccines, bacterial diseases, neglected tropical diseases, infectious disease control, human geography, research and analysis methods, hydrology, public and occupational health, infectious diseases, geography, cholera, natural resources, cholera vaccines, environmental engineering, earth sciences | null |
journal.pgen.1000319 | 2,009 | A Genome-Wide Association Study Identifies Novel and Functionally Related Susceptibility Loci for Kawasaki Disease | Kawasaki disease ( KD; MIM 611775 ) is an inflammatory vasculitis predominantly affecting young children 1 ., It is characterized by a striking propensity for coronary artery damage , which occurs in approximately 25% of untreated and 3–5% of treated children ., KD is the commonest cause of heart disease acquired in childhood in developed nations and in those who manifest coronary artery damage , KD may be associated with serious cardiovascular sequelae in adulthood 2 ., The long-term cardiovascular implications of KD in those without overt coronary artery lesions are unclear ., The etiology of KD is unknown , but it is thought to reflect an abnormal and sustained inflammatory response to one or more infectious triggers in genetically susceptible individuals 1 , 3 ., No consistent etiologic agent for KD has been identified , hampering accurate and timely diagnosis and the development of optimal management strategies ., The incidence of KD varies markedly in different ethnic groups , with the highest incidence in North East Asian populations ., KD affects approximately 1 in 150 Japanese children 1 and is responsible for 1–2% of all pediatric hospital admissions in South Korea 4 ., There are strong epidemiologic data to support a substantial genetic contribution to KD susceptibility ., The Japanese incidence ( 135–200/100 000<5 years of age ) is 10–15 times greater than the Caucasian incidence ( 9–17/100 000<5 years of age ) 1 and this difference is maintained in American children of Japanese descent resident in the US 5 ., Other Asian populations in the UK 6 , 7 and the US 8 also have a significantly higher incidence than non-Asians residing in the same geographic location ., The familial inheritance pattern of KD is in keeping with a polygenic complex disease and in multi-case pedigrees , KD occurs in family members at different times and geographic locations 9 ., Across all populations KD is approximately 1 . 6 times more common in males 3 ., The sibling risk ratio for KD in the Japanese is approximately 10 during an epidemic 10 and 6 overall 11 ., KD is over twice as common in the children of parents who themselves had KD in childhood , with multi-generational pedigrees often having more than one child affected , an earlier age of onset and a more severe phenotype 11 ., A genome-wide linkage study identified three regions exhibiting modest linkage in Japanese KD sibling pairs 12 ., Detailed association analysis of the linkage peak at chromosome 19q13 . 2 identified a significantly associated functional variant in ITPKC ( Gene ID:80271 ) , a negative regulator of T cell activation 13 ., In both Japanese and US Caucasian children with KD , this variant was associated with an approximate doubling of KD risk 13 ., Other investigations of KD susceptibility determinants to date have been candidate gene association studies ., A number of immunologic and cardiovascular-related loci have revealed genetic associations of modest effect size , but studies have often been under-powered and the findings have rarely been replicated in independent populations ., As these associated candidate loci are likely to account for only a small proportion of the overall genetic susceptibility to KD , we undertook a genome-wide association study ( GWAS ) to identify novel loci that might mediate susceptibility to KD ., We performed the initial GWAS in a well-characterized Dutch Caucasian population and tested the most significantly associated SNPs and haplotypes in a large independent cohort of predominantly Caucasian trios from three countries ., Fine-mapping of sixteen variants with a minor allele frequency ( MAF ) of >0 . 05 that lay within or close to known genes identified eight significantly associated variants , five of which may be functionally related ., We included 119 KD cases and 135 controls in the initial GWAS ( Table 1 ) ., Ten non-Caucasian subjects were excluded following admixture analysis by Eigenstrat 14 ., Three samples were excluded due to genotyping call rates <93% ., The final GWA analysis therefore consisted of 107 KD subjects and 134 controls ., Of the 262 , 264 SNPs on the Affymetrix 250 K NSP chip , 18 , 211 had a call rate of <93% , 18 , 981 were monomorphic ( MAF<0 . 1% ) and 1 , 150 deviated significantly from HWE in the control group ., 223 , 922 SNPs were therefore available for analysis , of which 5 , 571 were on the X chromosome ., A total of 14 , 065 SNPs were significantly associated ( p<0 . 05 ) ., The quantile-quantile plot between observed and expected allele frequencies showed deviation from expected with p<∼10−4 , suggesting the presence of true associations 15 ( Figure 1 ) ., We undertook a replication study in an independent cohort of KD cases and parental controls , using an exact replication strategy , genotyping only the most significantly associated variants by a different genotyping technology ., After verifying family relationships and checking sample duplications , 63 samples were excluded from further analysis ., Thus 1 , 903 members of 583 KD families , including 498 trios , were tested in the follow-up association analysis ., The 1 , 148 most significantly associated SNPs variants ( corresponding to a minimum significance level of p<0 . 0024 ) were selected by a combination of Armitage trend test , recessive-dominant and allelic association analysis ( Figure 2 ) ., SNPs were genotyped in the follow-up cohort by a custom Illumina Oligo Pool Assay , in which 1 , 116 SNPs were successfully genotyped ( Table S1 ) ., Fifteen SNPs failed quality filters , leaving a total of 1 , 101 SNPs ., Significant associations with KD susceptibility were replicated for 40 SNPs ( Table 2 ) ., Twenty eight lay either in or within 50 kb of known genes ., The most highly associated SNP was located in the gene for N-acetylated alpha-linked acidic dipeptidase-like 2 ( NAALADL2 , Gene ID:254827 ) , ( combined OR from case-control and family analyses\u200a=\u200a1 . 43 ( 1 . 32–1 . 53 ) ; pcombined\u200a=\u200a1 . 13×10−6 ) ., Three SNPs were located in introns of the gene for AT-binding transcription factor 1 ( ZFHX3 , Gene ID:463 ) , all of which had a protective effect ( ORcombined\u200a=\u200a0 . 64 ( 0 . 52–0 . 75 ) , 0 . 68 ( 0 . 57–0 . 79 ) , 0 . 73 ( 0 . 62–0 . 84 ) ; pcombined\u200a=\u200a2 . 37×10−6 , 7 . 06×10−5 , 5 . 35×10−4 , respectively ) ., Two of these SNPs , rs7199343 and rs10852516 , were only 647 bp apart and in high linkage disequilibrium ( LD ) ( r2\u200a=\u200a0 . 93 ) , whereas the third SNP , rs11075953 , was ∼10 kb distant and in less LD ( r2\u200a=\u200a0 . 46 ) ., Two SNPs , rs1010824 and rs6469101 , which were 325 bp apart and in complete LD ( r2\u200a=\u200a1 ) , were situated ∼10 kb 3′ of the angiopoietin 1 gene ( ANGPT1 , Gene ID:284 ) ( ORcombined\u200a=\u200a0 . 58 ( 0 . 43–0 . 73 ) and 0 . 6 ( 0 . 45–0 . 76 ) ; pcombined\u200a=\u200a3 . 39×10−5 and 5 . 44×10−5 , respectively ) ., 166 SNPs within the most highly associated thirty five haplotype blocks ( corresponding to p<2×10−3 ) were genotyped using a Sequenom iPLEX platform in the replication cohort ( Table S2 ) ., Twenty-nine SNPs could not be genotyped , leaving 137 SNPs for further analysis ., Significant associations were verified for six haplotypes , of which four were within known genes ., Two haplotypes lay within the gene encoding M-phase phosphoprotein 10 ( U3 small nucleolar ribonucleoprotein ) , MPHOSPH10 ( Gene ID:10199 ) ., One of these haplotypes was associated with increased susceptibility ( ORcombined\u200a=\u200a3 . 52 ( 1 . 6–7 . 75 ) ; p\u200a=\u200a3 . 69×10−3 ) and the other with protection from KD ( ORcombined\u200a=\u200a0 . 39 ( 0 . 24–0 . 63 ) ; p\u200a=\u200a3 . 91×10−4 ) ( Table 2 ) ., In order to narrow the region of association within selected genes , we carried out fine-mapping of sixteen of the genes replicated in the family study ( Table S3 ) ., We fine-mapped genes identified by variants with a MAF of >0 . 05 lying within 5 kb of known genes ., Polymorphisms from eight of the sixteen fine-mapped genes showed significant combined p-values ( Table 3 ) ., In three genes the tagging SNP showed the most significant genetic association; LNX1 ( Gene ID:84708 ) ( rs7660884 , pcombined\u200a=\u200a1 . 8×10−3 ) , CAMK2D ( Gene ID:817 ) ( rs11728021 , pcombined\u200a=\u200a1 . 3×10−2 ) and CSMD1 ( Gene ID:64478 ) ( rs2912272 , pcombined\u200a=\u200a3 . 5×10−2 ) , indicating that the initial GWAS probably identified a disease-associated haplotype ., Conversely within NAALADL2 , the most significantly associated gene identified in the replication study , a polymorphism ( rs1870740 ) located 23 kb away from the tagging SNP ( rs2861999 ) was the most significantly associated ( pcombined\u200a=\u200a9×10−4 ) ., A linkage disequilibrium plot of this region indicated that these two SNPs belong to distinct haplotype blocks ( Figure 3 ) ., The transcription factor ZFHX3 had a number of SNPs with more significant associations than the variant tagging the replicated polymorphism ( Figure S2 ) ., Linkage disequilibrium plots of all fine-mapped genes are presented in the online supplemental material ( Figure S3 , S4 , S5 , S6 , S7 , and S8 ) ., In order to investigate differences in allele frequencies of associated SNPs between Japanese , where the incidence of KD is approximately twenty times higher than that of Caucasians , we compared MAF data for associated variants using HapMap ., In addition we investigated the ancestral allele of each associated SNP from available data from higher primates ( Chimpanzee and Macaque ) ( Table S4 ) ., All ( but one ) associated alleles of SNPs within ZFHX3 had higher frequencies in the Japanese population ., The associated allele ( T ) from the most significantly associated SNP ( rs9937546 pcombined\u200a=\u200a1 . 9×10−4 ) , had an allele frequency of 0 . 922 in the Japanese , compared to 0 . 633 in Caucasians ., Despite the high frequencies in human populations , the rs9937546 T allele was not the ancestral allele , which might indicate rapid fixation in the population due to an unidentified evolutionary advantage ., In contrast the associated allele of rs2912272 from CSMD1 , was absent in Japanese populations , which could indicate genetic heterogeneity in this gene , with other variants associated with susceptibility in the Japanese ., In CSMD1 the ancestral allele is the major allele in humans ., We explored possible functional relationships between the eight genes confirmed by fine-mapping using the Ingenuity Pathway Analysis ( IPA ) Knowledge Base ., Unsupervised IPA network analysis identified a single cluster of 35 genes that included five of the eight associated genes and 26 additional genes , that was unlikely to occur by chance ( p\u200a=\u200a10−13 ) ., Highlights from the group are shown in Figure 4 , concentrating on the close connection between four of the associated fine-mapped genes and eight IPA identified genes that create a plausible biological network ., We investigated the whole blood transcript levels of the eight fine-mapped associated genes , and three of the IPA identified genes , using TaqMan quantitative PCR , in 27 patients with paired samples from their acute KD stage ( prior to treatment ) and their convalescence , using the transcript levels of the gene 18S as a loading control ., The blood cell profiles of these patients have been described previously 16 ., Of the 54 RNA samples , two pairs were excluded because of inadequate RNA quality ., Of the eight GWAS identified genes , five showed significantly lower abundance at the acute phase compared to convalescence , while two showed no change and expression of one was not detected in peripheral blood ( Table 4 ) ., For the three IPA identified genes investigated , one showed significantly higher abundance , one showed significantly lower abundance and one showed no significant difference between acute and convalescence samples ( Table 4 ) ., NAALADL2 ( NM_207015 ) showed the largest fold change ( FC ) ( FC\u200a=\u200a−6 . 3 , p\u200a=\u200a1 . 55×10−4 ) , while CAMK2D ( NM_001221 ) showed the most significant difference ( FC\u200a=\u200a−2 . 5 , p\u200a=\u200a1 . 53×10−7 ) ( Table 4 ) ., Relative transcript abundance is represented in the network diagram ( Figure 4 ) , with red indicating greater abundance in acute samples compared to convalescent samples , green lower abundance and yellow no change ., To our knowledge , this is among the first GWAS of an infectious disease and the first GWAS of KD ., We have identified a number of novel variants using a staged study design and subsequent fine-mapping that are associated with KD susceptibility ., These include variants within or close to genes that are functionally inter-related and that are plausible biological candidates in the KD pathogenesis ., The magnitude of the effect sizes for KD susceptibility is comparable to that reported from other GWAS 17 ., Fine-mapping of associated and replicated SNPs has focused on more frequent variants that lie in known genes ., In eight of these sixteen genes , fine-mapping confirmed the association and identified one or more associated haplotype ( s ) , which will form the basis of resequencing to identify the disease-modifying variants ., The assertion that these variants are in ( or close to ) biologically relevant loci involved in KD susceptibility is supported by;, ( i ) identification of eight loci containing one or more independently associated haplotypes identified by GWAS , replicated in an independent family-based study and subsequently fine-mapped ,, ( ii ) the significant differential gene mRNA transcript abundance of 5 of the 7 blood-expressed fine-mapped genes during acute versus convalescent KD , and, ( iii ) the gene network analyses that suggest biologically plausible functional relationships , which are extremely unlikely to have occurred by chance , exist between five of the associated loci ., We focused on fine-mapping of associated SNPs that lie either in or within 5 kb of known genes and had a MAF of >0 . 05 in HapMap ., These data represent the most robust associations and we will therefore focus our discussion on those genes , where putative functional relationships were suggested by IPA ., We used IPA in an unsupervised manner , allowing identification of gene-gene relationships without a priori assumptions ., This analysis linked five of the eight genetically associated genes , of which four form a functionally closely related network linked to eight other nodes in a highly significant network ., The gene network suggests possible mechanisms by which one or more infectious triggers may lead to dysregulated inflammation and apoptosis , and cardiovascular pathology ., Central to the putative gene network is CAMK2D ( calcium/calmodulin-dependent protein kinase ( CaM Kinase ) II delta ) , whose expression was significantly down regulated during acute KD ., CAMK2D encodes the δ-isoform of CaM kinase II ( NP_001212 ) , a ubiquitously expressed calcium sensitive serine/threonine kinase ., The δ-isoform of CaM kinase II is the predominant form expressed in cardiomyocytes and vascular endothelial cells 18 and is involved in a number of pathophysiological processes that make it an attractive candidate in KD ., In vascular endothelial cells CaM kinase II mediates nitric oxide ( NO ) production by endothelial synthase ( NOS3 , NP_000594 ) in response to changes in intracellular calcium and NO causes local vasodilatation 18 ., In acute KD NO production is increased and NO metabolites decrease following successful treatment 19 ., Following KD , especially where there has been overt CA damage , there is endothelial dysfunction and impaired vasodilatation , which can be restored after administration of antioxidants that may increase local availability of NO 20 ., More chronically , NOS3 may become dysregulated ( ‘uncoupled’ ) and produce potentially harmful superoxide anions , resulting in chronic oxidant stress that is implicated in the pathogenesis of atherosclerosis 21 ., In those with severe KD , NOS3 is expressed in coronary artery aneurysm tissue removed at surgery , and the tissue shows a pattern of senescence that is also typical of atherosclerosis 22 ., Involvement of leukocyte expressed CaM kinase II in blood vessel damage and aneurysm formation , key features of KD , is also plausible ., In human monocytes , CaM kinase II modulates tumor necrosis factor-induced expression of CD44 ( NP_000601 ) , which has a central role in leukocyte migration and extravasation at inflammatory sites 23 ., CaM kinase II is also involved in disruption of the endothelial barrier following stimulation with agonists such as thrombin 24 , whose levels may be increased following KD 25 ., Disruption of barrier integrity in coronary arteries may contribute to leukocyte infiltration into the vessel wall , proteolysis of extracellular matrix proteins and the internal elastic lamina and subsequent coronary artery aneurysm formation , that is pathognomonic of KD 1 ., In addition , CaM Kinase II regulates endotoxin- and TNF-mediated apoptosis in human promonocytic cells by regulating the anti-apoptotic gene BIRC3 ( Gene ID:330 ) 26 ., Delayed apoptosis of leukocytes is characteristic of acute KD and may contribute to pathogenesis 27 ., Intravenous immunoglobulin ( IVIg ) , standard therapy for KD , induces apoptosis of neutrophils in acute KD 28 ., In a genome-wide transcriptional study of KD , there was a marked over-representation of apoptosis regulatory genes 16 ., Both CaM Kinase II and the product encoded by another fine-mapped gene LNX1 , i . e . ligand of numb-protein X1 ( NP_116011 ) , interact with the NUMB family of proteins 29 , 30 ., Interestingly , one of the NUMB family members , NUMBL ( numb homolog ( Drosophila ) -like , Gene ID:9253 ) lies in the same small haplotype block that has recently been associated with KD susceptibility by a linkage study and subsequent fine-mapping 12 , 13 ., The NUMB gene ( Gene ID:8650 ) showed significantly higher transcript abundance during acute KD ., Both LNX1 and LNX2 ( Gene ID:222484 ) ( a closely related gene identified by the IPA network ) encode proteins that bind the coxsackievirus and adenovirus receptor ( CXADR , Gene ID:1525 ) 31 ., CXADR is the receptor for coxsackievirus B3 , which causes myocarditis in humans ., The myocarditis can be prevented in animal models by antagonizing viral binding to CXADR ( NP_001329 . 1 ) 32 ., Coxsackievirus B3 has also been implicated in acute myocardial infarction 33 ., Interestingly , the human endogenous retrovirus K protein Np9 also interacts with LNX1 34 and therefore a number of viruses may theoretically bind the NUMB/CAR/LNX1 complex , leading to internalization and regulation of CAMK2D activity ., This suggests a possible mechanism whereby more than one infectious trigger may result in cardiovascular damage in genetically susceptible individuals suffering from KD ., Other fine-mapped IPA-networked genes include ZFHX3 ( also known as ATBF1 ) , which encodes a large enhancer-binding transcription factor that is known to be polymorphic 35 and interacts with a number of proteins , including PIAS3 ( protein inhibitor of activated STAT , NP_006090 ) that inhibits STAT3 ( signal transducer and activator of transcription-3 , NP_644805 ) 36 ., STAT3 is activated by interleukin 6 ( IL6 , NP_000591 ) a pro-inflammatory cytokine that is involved in early innate immune reactivity , as indicated by the high fever , acute phase response with increased levels of CRP ( NP_000558 ) , complement factors and fibrinogen , in the blood as well as the myriad of cellular markers altered in acute KD 37 ., ZFHX3 also interacts with MYH7 ( myosin , heavy chain 7 , cardiac muscle , beta , NP_000248 ) , in which mutations are known to cause an inherited form of cardiomyopathy 38 ., CSMD1 ( CUB and Sushi multiple domains 1 ) , which is functionally related to CaM kinase II via histone deacetylase 4 ( HDAC4 , Gene ID:9759 ) , may be associated with dampening the early phase of KD ., CSMD1 is located on chromosome 8 , in a region that is hypervariable in humans and which contains numerous immune-related genes 39 ., Activation of the classical complement pathway occurs in acute KD 40 , and CSMD1 ( NP_150094 ) is a complement regulatory protein that blocks the classical but not alternate complement pathway 41 ., The functions of other fine-mapped genes are generally poorly understood ., The most significantly associated gene ( NAALADL2 , N-acetylated alpha-linked acidic dipeptidase-like 2 ) , which also showed the greatest change in transcript levels between acute and convalescent KD , is a large gene of 32 exons spanning 1 . 37 Mb ., NAALADL2 undergoes extensive alternative splicing leading to multiple 5′ and 3′ untranslated regions and variable coding sequences ., Its function is largely unknown , but mutations in the gene may contribute to Cornelia de Lange syndrome ( OMIM: 122470 ) 42 ., Overall we have identified five genetically associated genes that also had significantly reduced transcript levels during acute KD , including three that are closely functionally related ( Figure 4 ) , suggesting that these genes may act together ., This novel network may be distinct for KD as differences in transcript abundance in these genes have not been previously described as being part of a typical inflammatory response expressed in blood cells ., Pathogen-specific host responses , identified by relative transcript abundance in the blood have been described for other infectious diseases 43 ., Investigation of transcriptome abundance in whole blood rather than specific cell populations allows assessment of the entire peripheral blood transcriptome 43 and may be particularly informative in diseases such as KD , where an infectious trigger is implicated but remains unidentified 1 ., In a genome-wide gene expression study of KD , variation in neutrophil and lymphocyte numbers , characteristic of acute KD 44 , 45 , were thought to account for approximately half of the variation in transcript abundance during the course of the KD illness 16 ., Although we did not investigate individual cell populations in the current study , the data suggest that relative changes in transcripts reflect qualitative as well as quantitative differences ., Given the enrichment of the expression profile with immune-related genes ( selected on the basis of associated loci ) , the changes in mRNA may not be more numerous than those expected by chance and do not provide definitive proof for the gene-specific associations ., While the number of subjects in our expression study is large enough to identify overall trends in the host response during KD , we are unable to comment on expression-related allelic association , which will be investigated in future studies ., There is a suggestion from peripheral blood expression data in KD that ‘person-specific’ gene expression patterns , possibly reflective of underlying genetic variation , may be present 16 ., Further investigation of the relationship between genomic associations and gene expression will be undertaken , although clearly genetic variants may be significantly associated with disease without resulting in alterations in gene expression ., Our sample of 893 cases represents a large genetic KD cohort drawn from a single ethnic group ., KD shares many features of other infectious diseases of young children , including fever , rash and changes to the mucous membranes ., There is no diagnostic test and laboratory parameters individually have insufficient sensitivity or specificity for diagnosis 1 ., In all study cohorts we employed a conservative and widely accepted KD case definition in an attempt to maximize phenotypic homogeneity and diagnostic specificity ., The similar ethnicity and ascertainment of KD cases in all cohorts reduces the risk of spurious associations 15 ., Our methodological approach is consistent with current best practice recommendations in GWAS design and analysis , which are aimed to identify robust associations and reduce type 1 errors 15:, ( i ) the discovery and replication cohorts were recruited using very similar ascertainment techniques and drawn from predominantly Caucasian populations , with careful analysis to exclude cryptic population admixture in the discovery phase , which used a case-control design;, ( ii ) the variants selected for replication were predominantly selected using single-point analysis , although we employed other models , including haplotypic analysis to maximize the informativeness of the initial GWAS data;, ( iii ) we employed different genotyping technologies in each of the discovery , replication and fine-mapping stages to reduce spurious associations arising from genotyping errors;, ( iv ) we limited our replication genotyping solely to variants identified in the discovery phase , as additional fine-mapping around associated variants in the replication phase may increase spurious associations 46;, ( v ) we used a staged study design to avoid conservative correction for multiple statistical comparisons that might mask associations of moderate effect size in this modestly sized sample;, ( vi ) we present joint analysis of the discovery and replication data , rather than considering the replication data in isolation and;, ( vii ) we have fine-mapped variants with a MAF>0 . 05 which lie within or close to known genes ., We are aware that the genomic coverage and power of the discovery phase of the GWAS were limited and calculate that the initial GWAS had only approximately 50% power to detect an OR of 2 . 0 with alpha<0 . 05 ., Our relatively modest sample size reflects the difficulties in recruiting for a relatively rare disease in which the phenotype is defined clinically ., Our approach therefore aimed to reduce the risk of type I errors by ensuring that a large and independent replication cohort was included as part of the initial design , as we did not expect the associated variants to reach genome-wide significance , given the cohort size in the GWAS discovery phase 47 ., It was therefore expected that neither previously reported and credible candidate gene associations in KD , such as IL4 ( Gene ID:3565 ) 48 , VEGFA ( Gene ID:7422 ) 49 , 50 , CCR5 ( Gene ID:1234 ) 51 , and MBL2 ( Gene ID:4153 ) 52 , nor the recently reported ITPKC variant 12 were replicated by the GWAS ., Our study has failed to identify these and almost certainly other as yet unidentified variants that represent additional major determinants of KD susceptibility ., We have identified a number of novel associated SNPs , confirmed by fine-mapping , which lie within or close to previously unrecognized candidates for KD ., The effect sizes , independent verification in different populations , differential transcript abundance and network analyses all indicate that at least a proportion of these variants represent novel genetic risk factors for KD ., Some of the associated genes may interact to mediate the deleterious effects of infection-driven inflammation on the cardiovascular system ., Further characterization of the associated genes and their functional interactions may lead to the identification of novel diagnostic and therapeutic targets in KD and may be informative about early pathogenic processes in other cardiovascular diseases ., We used a staged study design with an initial GWAS , an exact replication phase in an independent cohort and subsequent fine-mapping of common variants lying within or near known genes ., We performed the initial GWAS analysis in a Dutch Caucasian case-control sample ( the ‘discovery phase’ ) and re-tested the most significantly associated SNPs and haplotypes in an independent sample of KD trios from Australia , the US and the UK , using a different genotyping technology ( the ‘replication phase’ ) ., Finally , a fine-mapping stage including sixteen replicated genes was performed in a subset of samples from discovery and replication phases , again using a different genotyping platform ., KD was defined by the presence of prolonged fever , together with at least four of the five classical diagnostic criteria 53 ., Children with at least five days of fever and two diagnostic criteria with echocardiographic changes of coronary artery damage during the acute and/or convalescent phases of KD were also included , as these coronary artery manifestations are pathognomonic for KD 53 ., Details of clinical symptoms of our study group can be found on Table S5 ., Cases of incomplete KD , who have fever , less than 4 diagnostic criteria and no coronary artery manifestations ( who constitute approximately 15% of KD cases receiving clinical treatment 54 ) , were excluded , to maximise the homogeneity of the clinical phenotype ., In all cohorts clinical and laboratory data were obtained directly from patient medical files and supplemented with parental questionnaires ., Phenotypic data were reviewed in all cases by experienced pediatricians ., Ethical approval was obtained from the appropriate national and regional institutional review boards for each study population ( Academic Medical Center AMC , Amsterdam ( Dutch cohort ) , UK Multi-Centre Research Ethics Committee ( UK cohort ) , each participating tertiary pediatric hospitals ethics committee ( Australian cohort ) and the University of California at San Diego ( US cohort ) ) ., Informed consent and assent as appropriate were obtained from participating families ., The initial GWAS was performed on 119 Dutch Caucasian KD cases and 135 healthy controls ., The cases were identified by collaborating pediatricians and sent for cardiological evaluation during the acute stage and subsequent follow-up to the AMC ., The controls were unrelated adult Caucasian blood donors residing in the same geographical area ., Ethnicity was determined by self or parental ethnic identification ., Assessment for possible population stratification was performed with Eigenstrat 14 ., Principal component analysis was applied to the genotype data to infer the axes of variation ., We used the top two principal components to identify outliers ., Any sample with principal component exceeding six standard deviations from the mean was identified as an outlier ., This process was repeated five times ., A lambda genomic control ( λGC ) , representation of stratification estimated after dividing the median ( chi-square ) by 0 . 456 , was calculated before and after running Eigenstrat ., We observed a λGC\u200a=\u200a1 . 18 before removal of potential outliers ., After running Eigenstrat ( sigma 6 . 0 , 5 iterations , n\u200a=\u200a10 individuals removed ) λGC dropped to 1 . 1 ., Dividing chi-square values by λGC we were able to correct for possible existence of population admixture ( Table 2 ) ., Sample duplication and family relationships were assessed by RelPair 55 ., A second independent cohort , which consisted of 583 KD-affected families , including complete and incomplete trios , from Australia , the US and the UK , was used to replicate the most significantly associated variants identified in the GWAS ., Family KD cases in each country were identified from pediatric hospital databases , through KD parent support groups and through media releases ., Biological parents ( where available ) and unaffected siblings ( to reconstruct missing parental genotypes ) were recruited ., A subset of samples from our GWA and follow up cohorts was genotyped in a fine-mapping experiment ., Due to limitations in DNA template we could include ∼85% of the original samples ., However , a new set of 493 samples were added in the case-control ( N\u200a=\u200a247 ) and family-based cohorts ( N\u200a=\u200a246 ) ., A principal component analysis comparing Hapmap populations with our cohort was applied to the genotype data of the case-control cohort to infer the axes of variation ( Figure S1 ) ., After removal of potential outliers ( N\u200a=\u200a113 ) λGC was 1 . 06 ., Allele chi-square values were divided by λGC and corrected p-values are reported on Table 4 ., Blood was collected | Introduction, Results, Discussion, Materials and Methods | Kawasaki disease ( KD ) is a pediatric vasculitis that damages the coronary arteries in 25% of untreated and approximately 5% of treated children ., Epidemiologic data suggest that KD is triggered by unidentified infection ( s ) in genetically susceptible children ., To investigate genetic determinants of KD susceptibility , we performed a genome-wide association study ( GWAS ) in 119 Caucasian KD cases and 135 matched controls with stringent correction for possible admixture , followed by replication in an independent cohort and subsequent fine-mapping , for a total of 893 KD cases plus population and family controls ., Significant associations of 40 SNPs and six haplotypes , identifying 31 genes , were replicated in an independent cohort of 583 predominantly Caucasian KD families , with NAALADL2 ( rs17531088 , pcombined\u200a=\u200a1 . 13×10−6 ) and ZFHX3 ( rs7199343 , pcombined\u200a=\u200a2 . 37×10−6 ) most significantly associated ., Sixteen associated variants with a minor allele frequency of >0 . 05 that lay within or close to known genes were fine-mapped with HapMap tagging SNPs in 781 KD cases , including 590 from the discovery and replication stages ., Original or tagging SNPs in eight of these genes replicated the original findings , with seven genes having further significant markers in adjacent regions ., In four genes ( ZFHX3 , NAALADL2 , PPP1R14C , and TCP1 ) , the neighboring markers were more significantly associated than the originally associated variants ., Investigation of functional relationships between the eight fine-mapped genes using Ingenuity Pathway Analysis identified a single functional network ( p\u200a=\u200a10−13 ) containing five fine-mapped genes—LNX1 , CAMK2D , ZFHX3 , CSMD1 , and TCP1—with functional relationships potentially related to inflammation , apoptosis , and cardiovascular pathology ., Pair-wise blood transcript levels were measured during acute and convalescent KD for all fine-mapped genes , revealing a consistent trend of significantly reduced transcript levels prior to treatment ., This is one of the first GWAS in an infectious disease ., We have identified novel , plausible , and functionally related variants associated with KD susceptibility that may also be relevant to other cardiovascular diseases . | Kawasaki disease is an inflammatory pediatric condition that damages the coronary arteries in a quarter of untreated patients and is the commonest cause of childhood acquired heart disease in developed countries ., While the infectious trigger ( s ) remain unknown , epidemiologic evidence suggests that human genetic variation underlies the susceptibility ., In order to identify novel mechanisms that may predispose to this disease , we undertook a genome-wide association study , which investigates genetic determinants without prior supposition regarding the loci of interest ., This was amongst the first complex infectious diseases to be studied in this way and one of the largest genetic studies of Kawasaki disease with 893 cases ., We identified and confirmed 40 SNPs and six haplotypes , identifying 31 genes , in an international cohort of Caucasian patients ., We followed up 16 SNPs where the associated genetic variant was more common and was situated within a gene , confirming eight SNPs by fine-mapping across the entire gene ., Of these eight genes , seven were expressed in blood and five showed significantly different gene expression in paired patient samples taken during acute and convalescent Kawasaki disease ., Five of the eight genes also appear to be involved in a single putative functional network of interacting genes ., These novel genes and pathways may ultimately lead to novel diagnostics and treatment for Kawasaki disease . | genetics and genomics/complex traits, infectious diseases | null |
journal.pcbi.1002032 | 2,011 | How Molecular Motors Are Arranged on a Cargo Is Important for Vesicular Transport | Cells are highly organized , and much of this organization results from motors that move cargos along microtubules ., The single-molecule properties of molecular motors are relatively well understood both experimentally and theoretically ., With this as a starting point , we investigated how the presence of the cargo itself alters transport ., Aside from exerting viscous drag , the cargo could in principle alter single-motor based transport both by changing the motors diffusion and ability to contact the filament ( a free motor diffuses very differently from a cargo-bound one ) , and also by exposing the motor to the random forces resulting from thermal fluctuations of the cargo which depend on the size of the cargo and the viscosity of the environment ., Whether such effects are significant are investigated here ., Recent studies show that cargos in vivo are frequently moved by more than one microtubule-based motor 1 , 2 , 3 , 4 ., This raises the question of how multiple motors function together , the subject of recent theoretical and experimental work 1 , 5 , 6 , 7 ., In vitro , when more than one motor is actively hauling a cargo , the run length , i . e . , the distance that the cargo travels along the microtubule before detaching , increases with the number of active motors ., However , the presence of the cargo itself may be important when there are multiple motors ., In addition to possibly changing the single-molecules function , the cargos size may alter the relationship between the total number of motors present and the number of motors actively engaged in transporting the cargo ( assuming random motor organization on the cargos surface ) ., If motors are not randomly organized , details of this organization will also be important ., How each of these factors contributes to overall transport is unknown ., To approach these problems requires a new theoretical framework: past studies simplified the problem using essentially one-dimensional models 5 , 6 , 8 , 9 that had the motors attached to the cargo at a single point , with the cargo represented by a single point ( though potentially experiencing viscous drag proportional to a specific diameter ) ., Here we have developed a bone-fide three dimensional Monte Carlo simulation that allows us to directly investigate how the presence of the cargo itself affects single-motor driven transport and motor-microtubule attachment , as well as how the relationship between cargo size and the arrangement of motors on the cargo affects ultimate cargo motion , all within the context of a cargo experiencing random Brownian translational and rotational motion ., The attachment of motors to a cargo of finite size , rather than an idealized point mass , has a number of ramifications ., First , the function of the motor ( s ) might be altered by the translational and rotational diffusion of the cargo; the larger the cargo , the more effect it has on the motors diffusion , and thus , potentially , on the motors ability to contact/interact with a microtubule ., Second , when a motor is attached to both the microtubule and the cargo , it will feel instantaneous forces due to the cargos thermal motion ., These forces will depend on the cargos size; and the random thermal ‘tugs’ from the cargo could slow the rate of travel of a motor and , in principle , induce the motor to detach from the filament ., Third , there is a relationship between the cargo size , the total number of motors present , how they are arranged , and how many can be engaged ., To illustrate this , imagine one cargo that is 50 nm in diameter , and another that is 500 nm in diameter ., In the first case , even if the motors are randomly distributed on the cargo , because the length of an individual motor is more than 100 nm , all of those on the lower half of the cargo , and some on the upper half , will be able to reach a nearby microtubule ( Figure 1A ) ., In contrast for the 500 nm cargo , most motors will be unable to reach if they are randomly distributed on the cargo ( Figure 1B ) ., However , if all the motors were clumped at a single point , the size of the cargo essentially becomes irrelevant , because if one motor can reach , they all can ( Figure 1C ) ., We thus set out to answer the following questions: We organized the presentation of our results according to these questions ., To address these questions , we developed three-dimensional Monte Carlo simulations ., Generally speaking , Monte Carlo is an approach to computer simulations in which an event A occurs with a certain probability PA where 0≤PA≤1 ., In practice , during each time step , a random number x is generated with uniform probability between 0 and, 1 . If x≤PA , event A occurs; if x>PA , event A does not occur ., Our simulations were carried out as follows ., We started with a three dimensional spherical cargo , subject to rotational and translational diffusion according to the equations presented below and in the Text S1 ., To this cargo , we attached kinesin motor ( s ) that are modeled as bungee cords , i . e . , they behave as springs with a spring constant of 0 . 32 pN/nm 5 , 10 when stretched beyond their relaxed length of 110 nm but produce no force when compressed ., We started the simulation so that potentially one or more motors could bind to a cylindrical microtubule ( 25 nm diameter ) ., The motors then moved the cargo along the microtubules , taking 8 nm steps ., While technical details of the simulation are in the Text S1 , the general idea is that at each time step Δt , we consider all motors present , calculate all forces acting upon them , and then ask what each of them does ., We start by describing how we simulate transport of a cargo with motors attached ., Our basic algorithm is as follows ., Consider one or more motors attached at random points to the cargo surface ., The cargo is then suspended above the microtubule , with a well-defined separation distance between the bottom of the cargo and the top of the microtubule , and the motors are each given an opportunity to attach to the microtubule ., If none do ( either because none can reach , or because although they can reach , they stochastically are not able to attach in the allotted time with the ‘on’ rate assumed to be ∼2/sec 11 , 12 , 13 ) , we use one of two initial conditions ., If we want to find the time it takes for a cargo with a single motor to attach , then the cargo is allowed to rotate consistent with Brownian diffusion , and the procedure is repeated ., Eventually , the motor binds ., The time between when the simulation is started and when the motor attaches is the ‘on’ rate for the cargo; since only one motor is present , it reflects how the presence of the cargo affects the motors on-rate ., The other initial condition is used if there are multiple motors and we are more interested in transport along the microtubule after the motors attach to the filament ., In this case , if none of the motors attaches after being given the opportunity to do so , the cargo is rotated so that at least one motor attaches to the microtubule ., Once some subset of the motors is attached , the cargo travels along the microtubule ., At each time step of the simulation , each motor on the cargo is given the opportunity to detach from the MT if it is attached , or attach if it is detached ( and geometrically can reach the MT ) ., If a motor is attached to a MT , then there is some probability that it will bind and hydrolyze ATP , and subsequently take a step ., Although kinesin is a two headed motor , we model each motor by a single kinesin head that hydrolyzes ATP in such a way that Michaelis-Menten kinetics is obeyed ., The probabilities of a motor detaching from the MT , releasing ATP , and taking a step are all dependent on the load on the cargo because the cargo exerts force on the motors ( see Text S1 ., This load has contributions from the externally applied force , the other motors which are pulling the cargo , and from thermal fluctuations ., The thermal fluctuations randomly rotate and translate the cargo which , in turn , can stretch the motor linkage and exert a load on the motor ., ( See below for further details on thermal fluctuations . ), Once all the motors have been given a chance to step , the cargo is translated and rotated according to the force and torque to which it is subjected ., The cargo travels along the microtubule until all the motors detach from the microtubule , and the ‘run’ ends; this then determines the run length of the cargo ., The velocity is calculated by dividing the distance the cargo moves by the travel time τ , where τ is typically 1 msec but may be as long as 10 msec ., Averaging over these velocities gives the average velocity ., To get good statistics , we simulate a specified number of runs with the same initial conditions to get a set of runs ., We also simulate a number of sets with different initial conditions to obtain good statistics ., In our simulations , the spherical cargo is subjected to thermal fluctuations which we can divide into translational and rotational components ., The equation of the cargos translational motion is given by the Langevin equation: ( 1 . 1 ) where m is the cargos mass and is the cargos velocity ., The drag force on the cargo is proportional to its velocity with the drag coefficient , where R is the cargos radius and is the coefficient of viscosity which is the kinematic viscosity multiplied by the specific gravity of the fluid ., is the sum of the forces due to an external force of magnitude FL and the force of the engaged motors pulling on the cargo ., We solve this equation in the Text S1 , and quote the solution here for the position of the cargo at time step t+Δt: ( 1 . 2 ) where is the standard deviation of a normal distribution and is a vector in Cartesian coordinates of the laboratory frame of reference that represents three independent random variates drawn on a normal distribution having zero mean and unit standard deviation ., For the cargos rotational motion , the corresponding Langevin equation is ( 1 . 3 ) where is the moment of inertia of a solid spherical cargo , and is the drag coefficient proportional to the angular velocity ., is the torque on the cargo referenced from the center of mass due to the engaged motors ., is the rapidly varying random torque due to the thermal fluctuations of the environment ., We solve this equation in the Text S1 where we give the formulas for the change in orientation of the cargo at each time step ., These formulas are analogous to Eq ., ( 1 . 2 ) ., As we shall see , rotational diffusion due to thermal fluctuations can play a significant role in limiting the distance that a cargo can travel ., After considering motors randomly attached anywhere on the cargo , we consider cases which have a restricted region of the cargo surface area where motors can attach ., For these cases , we start each simulation with N motors randomly attached to the cargos surface within a region specified by the cone angle as shown in Figure, 2 . The area available for attachment can be described by a cone with its apex at the center of the sphere ., A line extends from the apex to the base of the cone ., The cluster angle φ is the angle between this line and the side of the cone ., The intersection of the cone with the surface of the cargo defines the allowed region of motor attachment ., The cluster angle can vary between 0 and 180 degrees ., A cluster angle of 90 degrees defines the lower hemisphere of the cargo ., A cluster angle of 180 degrees corresponds to the entire spherical surface , and means that the motors can attach anywhere on the sphere ., Our study of the effects of the cargo on transport has a number of ‘take-home’ messages ., The first is that , at both the single-motor and multiple-motor levels , the presence of the cargo can significantly alter the effective ‘on’ rate/probability of successful binding of the motor ( s ) to the filament , because the center of mass of the cargo diffuses away from the microtubule relatively slowly , and while this is occurring , its rotational diffusion frequently brings the motor close enough to the microtubule to allow attachment ., Thus , the cargo ‘helps’ the motor to attach , though the degree of assistance depends on cargo size and viscosity of the medium surrounding the cargo ., Rapidly diffusing cargos might not linger long in the vicinity of a microtubule , but in a cell where there are multiple filaments available , these cargos could quickly find and bind to a filament ., Second , in order to for a motor to attach to the filament in a reasonable amount of time , the motor length needs to be longer or comparable to the radius of the cargo which may explain why motors are 60 to 110 nm in length ., Third , if motors are randomly arranged on the cargos surface , the relationship between the number of motors present and the number of actually engaged motors depends strongly on the cargo size , so that different simple models of regulating cargo motion by recruiting motors to the cargo surface ( either by a specified change in total number of motors , or by a specified change in local motor surface density ) will have different effects on overall cargo motion as a function of cargo size ., Thus , in order to have regulation affect a set of cargos equally , independent in variations in cargo size , it is best to have motors clustered in a small region on the cargo ., A further finding also supports the utility of motor clustering: for large cargos , if motors are randomly placed , achieving a reasonable number of engaged motors ( n\u200a=\u200a3–6 ) would require a large number of motors ( 50–100 ) to be present on the cargo , which appears inconsistent with biochemical characterizations of cargo-bound microtubule motors 4 , though it is consistent with biochemical characterizations of cargo-bound myosin motors 18 which are likely randomly arranged on cargos 18 , 19 ., Overall , our findings suggest that , in vivo , microtubule motors are likely organized into clusters when present on large cargos , but that such clustering is unnecessary for small cargos ., In addition , a reasonable number of engaged motors would be required for long travel distances of several microns but not for short run lengths ., Since microtubules can be tens of microns long compared to actin filaments which have a typical decay length of 1 . 6 microns 18 , we expect long travel distances along microtubules but relatively short run lengths along actin filaments ., Thus we predict the microtubule motors kinesin and dynein to be clustered on cargos while we expect the actin motor myosin V to bind randomly to cargos ., There is clear experimental evidence for the random arrangement of myosin on cargos in vivo , and weak experimental evidence for the clustering of kinesin and cytoplasmic dynein 19 ., For the purposes of this paper , we have assumed that the points where motors are attached to the cargos are fixed on the cargos surface ., This is true in some cases , e . g . , when motors bind to dynactin which in turn binds to spectrin which is a filament that coats some vesicles 20 , 21 ., However , in other cases , the attachment points can diffuse through the fluid membrane of the vesicle and cluster at one location ., An example of this is an experiment showing that motors dynamically accumulate at the tip of membrane tubes growing out of a vesicle as a consequence of the fluidity of the membrane 13 , 22 ., Clustering does not seem to affect the rate at which the first motor of a cargo attaches to a microtubule unless the cargo is large ( greater than 200 nm ) and the viscosity is high ., Motor proteins are sufficiently long ( greater than 50 nm ) and rotational diffusion sufficiently rapid that the number of motors on a cargo does not significantly affect the rate at which the cargo binds to the microtubule . | Introduction, Methods, Discussion | The spatial organization of the cell depends upon intracellular trafficking of cargos hauled along microtubules and actin filaments by the molecular motor proteins kinesin , dynein , and myosin ., Although much is known about how single motors function , there is significant evidence that cargos in vivo are carried by multiple motors ., While some aspects of multiple motor function have received attention , how the cargo itself —and motor organization on the cargo—affects transport has not been considered ., To address this , we have developed a three-dimensional Monte Carlo simulation of motors transporting a spherical cargo , subject to thermal fluctuations that produce both rotational and translational diffusion ., We found that these fluctuations could exert a load on the motor ( s ) , significantly decreasing the mean travel distance and velocity of large cargos , especially at large viscosities ., In addition , the presence of the cargo could dramatically help the motor to bind productively to the microtubule: the relatively slow translational and rotational diffusion of moderately sized cargos gave the motors ample opportunity to bind to a microtubule before the motor/cargo ensemble diffuses out of range of that microtubule ., For rapidly diffusing cargos , the probability of their binding to a microtubule was high if there were nearby microtubules that they could easily reach by translational diffusion ., Our simulations found that one reason why motors may be approximately 100 nm long is to improve their ‘on’ rates when attached to comparably sized cargos ., Finally , our results suggested that to efficiently regulate the number of active motors , motors should be clustered together rather than spread randomly over the surface of the cargo ., While our simulation uses the specific parameters for kinesin , these effects result from generic properties of the motors , cargos , and filaments , so they should apply to other motors as well . | The spatial organization of living cells depends upon a transportation system consisting of molecular motor proteins that act like porters carrying cargos along filaments that are analogous to roads ., The breakdown of this transportation system has been associated with neurodegenerative diseases such as Alzheimers and Huntingtons disease ., In living cells , cargos are typically carried by multiple motors ., While some aspects of multiple motor function have received attention , how the cargo itself affects transport has not been considered ., To address this , we developed a three-dimensional computer simulation of motors transporting a spherical cargo subject to fluctuations produced when small molecules in the intracellular environment buffet the cargo ., These fluctuations can cause the cargo to pull on the motors , slowing them down and making them detach from the filament ( road ) ., This effect increases as the cargo size and viscosity of the medium increase ., We also found that the presence of the cargo helped the motors to bind to a filament before it drifted away ., If other filaments were present , then the cargo could bind to one of them ., Our results also indicated that it is better to group the motors on the cargo rather than spread them randomly over the surface . | physics, biophysic al simulations, biophysics theory, biology, computational biology, biophysics simulations, biophysics | null |
journal.pgen.1007588 | 2,018 | Numerous recursive sites contribute to accuracy of splicing in long introns in flies | RNA splicing is a crucial step in the mRNA lifecycle , during which pre-mRNA transcripts are processed into mature transcripts by the excision of intronic sequences ., Introns are normally excised as a single lariat unit ., However , some introns in the Drosophila melanogaster genome are known to undergo recursive splicing , in which two or more adjacent sections of an intron are excised in separate splicing reactions , each producing a distinct lariat 1 , 2 ., Recursively spliced segments are bounded at one or both ends by recursive sites , which consist of juxtaposed 3 and 5 splice site motifs around a central AG/GT motif ( with “/” indicating the splice junction ) 1 , 3 ., This mechanism appears to be restricted to very long Drosophila introns 3 , 4 ., However , because recursive splicing yields an exon ligation product identical to that which would have been produced from excision of the intron in one step , the genome-wide prevalence and function of recursive splicing have been difficult to ascertain 3 , 4 ., Recursive splicing was initially observed in the splicing of a 73 kb intron in the Drosophila Ultrabithorax ( Ubx ) gene , where the intron is removed in four steps through intermediate splicing of the 5 splice site to two microexons and one recursive site before pairing with the proper 3 splice site 1 ., Bioinformatic searches for recursive sites predicted a couple hundred possible recursive sites in Drosophila , predominantly in introns larger than 10 kb 3 , but sites in only four introns , all from developmentally important genes ( Ubx , kuzbanian ( kuz ) , outspread ( osp ) , and frizzled ( fz ) ) , could be experimentally validated 1–3 ., Biochemical characterization showed that recursive splicing is the predominant processing pathway for splicing of these introns , which are generally constitutively spliced 1–4 ., More recently , an analysis by Duff and coworkers of all ~10 billion RNA-seq reads generated by the Drosophila ModENCODE project identified 130 recursively spliced introns in flies 4 ., Using this larger catalog of recursive sites , they confirmed that recursive splicing is a conserved mechanism to excise constitutive introns , requires canonical splicing machinery , and only occurs in the longest 3% of Drosophila introns 4 ., Similar analyses of mammalian RNA-seq datasets have resulted in the identification of just a handful of recursively spliced introns , mostly in genes involved in brain development , despite the greater abundance of long introns in vertebrate genomes 5 ., The scarcity of validated examples suggests that recursive splicing is quite rare , even in Drosophila ., However , the transient nature of recursive splicing intermediates makes it difficult to detect evidence for recursive splicing using standard RNA-seq data ., Support for recursive splicing has come from RNA-seq reads that span a junction between a known splice site and a putative recursive splice site internal to an intron , or from observation of a sawtooth pattern of reads resulting from the splicing out of recursive segments 4 , 5 ., Previous studies using polyA-selected RNA-seq data– which derive predominantly from mature transcripts– had limited ability to detect such evidence ., However , nascent RNA sequencing , which profiles pre-mRNA transcripts shortly after they are transcribed , should enable much more efficient capture of reads from intermediates of splicing , including recursive splicing ., Using such data should allow for more unbiased and systematic discovery of recursive splicing ., To globally detect transient splicing intermediates indicative of recursive splicing , we applied novel computational approaches to high-throughput sequencing data from short time period metabolic labeling of RNA ., This approach detected about four times as much recursive splicing as had been previously observed ., This expanded catalog of sites and associated analyses suggests a function for recursive splicing in improving splicing accuracy ., Pre-mRNA splicing can initiate immediately after transcription of an intron is completed , and can occur in as short a time as one or a few seconds 6–9 ., Since recursive splicing involves the splicing of intermediate intronic segments , it may begin soon after the transcription of the first intronic recursive site ., Thus , to have the greatest chance of capturing recursive splicing intermediates , it is essential to capture nascent transcripts as soon as possible after transcription , before introns have been fully spliced ., Here , we used nascent RNA sequencing data from our recent study , which used incorporation of a metabolic label to isolate RNA at short time points after transcription 9 ., The experimental approach to collect these data involved 5 , 10 , or 20 min labeling with 4-thiouridine ( 4sU ) in Drosophila S2 cells and 4sU biotinylation to selectively isolate nascent RNA , followed by RNA sequencing with paired-end 51 nt reads 9 ., These data were complemented by steady state RNA-seq data representing predominantly mature mRNA ( Methods ) ., The progressive labeling strategy used for these data results in isolation of transcripts that initiated during the labeling period , in addition to transcripts that were elongated during this period but initiated prior to the addition of the label 9 ., While this likely does not significantly bias the distribution of fragment lengths sequenced , there is an overall 5 to 3 bias of reads across the entire transcript ., We hypothesized that this high-coverage nascent RNA data would more readily identify recursive sites and better characterize the prevalence of recursive splicing ., For this purpose , we used a computational pipeline to detect three key signatures of recursive splice sites ( Fig 1 ) ., First , we used a custom python script to search for splice junction reads derived from putative recursive sites ( RatchetJunctions ) , as previously described ( Methods; Fig 1A ) 4 , 5 ., Ratchet junction reads contain a segment adjacent to an annotated 5 or 3 splice site juxtaposed to a segment adjacent to an unannotated intronic recursive site , providing direct evidence for the presence of a recursive splicing event ., Second , we developed a new computational tool , RatchetPair , to identify read pairs that map to distant genomic sites in a manner such that presence of intervening recursive splicing can be inferred from the size distribution of inserts in the sequenced library ( Methods; Fig 1A ) ., Unlike ratchet junction reads , recursive junction spanning read pairs do not pinpoint a specific recursive site ., Instead , a recursive site is inferred based on the empirical distribution of fragment lengths and genomic sequence information ., To do so , we adapted the GEM algorithm 10 , originally designed to infer protein binding sites from ChIP-seq data , to assign a probability that each read-pair was indicative of a recursive site in a given region ( Methods ) ., This modified GEM algorithm was run with all read pairs and splice junction reads pooled together to derive the empirical distribution of fragment lengths ., Third , we developed the first automated software , RatchetScan , for inference of recursive sites from sawtooth patterns in read density ( Fig 1B ) ., This type of pattern is an expected product of co-transcriptional recursive splicing and has been associated with many recursive introns 4 , 5 , 11 ., Briefly , assuming that RNA is spliced shortly after transcription elongation past the recursive splice site or 3 splice site , the splicing of recursive segments during transcription of subsequent sequences will result in a sawtooth distribution of reads across the intron with recursive sites commonly located near the right-hand base of each “tooth” ., It is important to note that these approaches do not differentiate between unproductive splicing ( followed by degradation of the intron-containing transcript ) and productive splicing of the full intron ., RachetScan predicts the locations of recursive sites in three distinct steps ., First , RNA-seq data was processed to summarize read density in each sub-intronic region ( S1A Fig ) ., We then developed a Markov Chain Monte Carlo- ( MCMC- ) based inference algorithm to detect presence of sawtooth patterns in introns ., This algorithm is suitable for efficient exploration of complex intronic read patterns encountered when considering a variable number of possible recursive splice sites in each intron ., We considered all nucleotides as potential recursive splice sites , rather than only focus on sites at the center of strong juxtaposed recursive motifs , allowing us to independently use sequence information to assess the false-positive rate of our method ., Our RachetScan algorithm is initiated with a randomly chosen state , consisting of a set of proposed recursive sites in the intron ( S1B Fig ) ., In each round , a new state is proposed by perturbing the current state , with three classes of perturbations: ( 1 ) a new recursive site is added; ( 2 ) a recursive site is removed; or ( 3 ) a recursive site location is locally shifted , each with defined probabilities ., Using a scoring function and transition rules ( detailed in Methods ) , the algorithm decides to either accept the new proposed state or maintain the current state ., This procedure was iterated over 107 rounds and the current state was sampled every 50 rounds , where the number of samples recorded in each state is proportional to the probability that the intron is best fit by the model corresponding to that state ., Finally , recursive sites are predicted based on the output of the inference algorithm and sequence information ( S1C Fig; S2 Fig ) ., This approach does not infer the order of splicing of recursive segments ( but see below ) ., Combining these three approaches and using reads pooled across all replicates and labeling periods , our analysis detected 539 candidate recursive sites in 379 fly introns ( S1 Table ) ., From this set , we curated a set of 243 “high confidence” recursive sites in 157 introns ( identified by at least 2 methods with greater than 5 supporting junctions or read pairs or a sawtooth FDR of 5% and visual inspection of read densities ) , and a “medium confidence” set of 296 sites ( identified by at least 1 method with greater than 5 supporting junctions or read pairs or a sawtooth FDR of 20%; Fig 2A; Methods ) ., Approximately 60% of our high-confidence sites ( 144 sites ) were identified using all three approaches ., Overall , 98 introns contained multiple recursive sites , with up to seven high-confidence sites observed in a single intron ., For instance , intron 1 of the tenascin major ( Ten-m ) gene contains five recursive sites , two of which were previously unknown ( Fig 1C ) ., Of the recursive sites previously reported by Duff and colleagues , 124 occurred in genes expressed in S2 cells ., Our approach detected 119 ( 96% ) of these known sites , as well as 126 novel high confidence sites and 294 novel medium confidence sites ( Fig 2A ) , thus increasing the number of recursive sites defined in this cell type by ~4-fold ( S2C Fig ) ., For three recursive segments , we were also able to detect reads that spanned the intronic lariat resulting from the second step of splicing ( S2 Table; Methods ) ., For 13 sites in 3 recursively spliced introns , we performed RT-PCR validation experiments using primers flanking recursive segments , followed by sequencing ( Methods ) ., These experiments validated 8 previously identified sites and 5 novel recursive sites in nascent RNA from Drosophila S2 cells , including 3 sites in an ~55 kb intron of Tet that was not previously known to be recursively spliced ( S3A Fig , S3D Fig; S3 Table ) ., Both the high confidence and the medium confidence candidate recursive sites exhibited a strong juxtaposed 3/5 splice site motif ( S4 Fig ) ., The greater numbers of sites detected by our approach ( 2–4 times more sites in this cell type ) , using less than 1/20th as many reads as used by Duff and colleagues , affirms the potential of nascent RNA analysis for identification of recursive splice sites ., Using this updated catalog of recursive sites , we observed that many very long introns ( > 40 kb in length ) have recursive sites , with 63% of such introns containing at least one high-confidence recursive site , and an additional 7% containing medium-confidence site ( s ) ( Fig 2B ) ., This observation suggests that recursive splicing is the prevalent mechanism by which very large fly introns are excised ., We assessed the sensitivity of our detection pipeline by running it on subsamples of reads ranging from 0 . 1% to 100% of the total reads ( Fig 2C ) ., The shape of the resulting curve tapered off at higher coverage levels but never plateaued: new recursive sites were still being detected as read depth increased from 50% to 100% of sequenced reads and therefore would likely increase further at higher read depths ., A somewhat higher proportion of recursive sites were detected in high-expressed genes ( TPM > 20 ) than low-expressed genes ( TPM ≤ 20 ) ., However , subsampling of the reads mapping to high-expressed genes to levels comparable to those observed for low-expressed genes resulted in a substantially lower fraction of recursive sites at each depth , suggesting that recursive splicing is more prevalent in low-expressed than high-expressed genes ( Fig 2C ) ., Together , these data suggest that the true fraction of very long introns that contain recursive sites may be substantially higher than our observed fraction of 63–70% , i . e . that recursive splicing is likely present in almost all very long fly introns ., Recursive splice sites can be required for the processing of long introns 3 ., However , it is possible that most recursive sites are functionally neutral , and that mRNA production is not impacted by their presence ., The size of our dataset enabled us to examine four properties of recursive sites that could help to distinguish between these possibilities: sequence conservation; distribution in the fly genome; distribution within introns; and efficiency of splicing ., In each case , the patterns observed suggest that recursive sites often have functional impact ., Both high and medium confidence recursive sites exhibited twice the level of evolutionary conservation observed in and around control AGGT motifs in long introns ( Fig 3A ) , implying strong selection to maintain most or all of these sites ., Recursively spliced introns were enriched in genes involved in functions related to development , morphogenesis , organismal , and cellular processes , with stronger enrichments observed for genes containing high-confidence recursive sites ( Fig 3B; S4 Table ) ., Both of these observations are consistent with results from a previous study based on a smaller sample of recursive introns 4 ., Longer introns might contain more recursive sites purely by chance ., Indeed , while the majority of recursively spliced introns had just one recursive site , the number of sites increased roughly linearly with intron length ( Fig 3C ) ., However , the positioning of recursive sites within introns was significantly biased away from a random ( uniform ) distribution ., Instead , recursive sites in introns with only one such site tended to be located closer to the midpoint of the intron than expected by chance ( Kolmogorov-Smirnov P = 0 . 003; Fig 3D ) ., Furthermore , the first recursive site in introns with two or three such sites tended to be located approximately 33% and 25% of the way from the 5 end of the intron , respectively ( Fig 3E ) ., The distribution of recursive sites within introns suggests that they are positioned so as to break larger introns into “bite-sized” chunks of intermediate size ( typically ~9–15 kb in length; S5A and S5B Fig ) rather than at random locations that would more often produce much longer and much shorter segments ., Recursively spliced introns were also enriched in first introns , which are longer than non-first introns , relative to subsequent introns in fly genes ( hypergeometric P < 0 . 05 ) ., To ask whether recursive splicing contributes to the efficiency of processing of very long introns , we evaluated the order and timing of recursive splicing events ( Methods ) ., We observed a steady increase in the proportion of exon-exon junction reads relative to recursive junctions across the time course , reflecting the progress of splicing ( Fig 4A ) ., Among recursive junction reads , we observed far higher counts of reads spanning the 5 splice site and the recursive site ( RS ) , relative to RS-RS or RS-3 splice site junctions , consistent with recursive segments being predominantly excised in 5 to 3 order ( Fig 4A; S5C Fig ) ., This order of splicing is consistent with recursive splicing occurring co-transcriptionally ., Using targeted RT-PCR amplification of segment combinations in nascent RNA from 3 recursively spliced introns , we were only able to detect products spanning the 5 splice site and recursive site ( S3B Fig ) ., Surprisingly , we did detect a product spanning the recursive site and 3 splice site for the third recursive site of the Ten-m intron in steady-state cDNA ( S3C Fig; validated by sequencing , S3D Fig ) , indicating that splicing of downstream recursive segments can sometimes occur before splicing of initial segments ., Finally , we also detected one read that spans a lariat resulting from a RS-RS junction ( S2 Table ) , as well as four reads ( for three junctions ) that span lariats resulting from the excision of recursive introns in one segment ( 5-3 junction ) ., The observation of these reads indicates that these introns are not always recursively spliced , though we note that these lariats are from introns that are much shorter than typical recursive introns ( 1 . 7–2 . 5 kb ) ., Previously we developed a framework for estimating rates of splicing from nascent RNA sequencing data across different labeling periods 9 ., Here , we adapted this approach to estimate the splicing half-lives of individual recursive segments ( Methods; S5D Fig; S5 Table ) , which have a mean length of 9 . 1 kb ( S5A Fig ) ., Recursive segment half-lives were the slowest for the first segment in the intron , with faster half-lives for successive segments ( S5E Fig ) ., Overall , recursive segments had 1 . 5-fold longer half-lives than non-recursive introns of the same lengths ( Fig 4B; Mann-Whitney P = 1 . 5×10−9 ) ., Estimating the mean splicing half-life of a recursive intron as the maximum of a set of exponentials ( to approximate the waiting time to splice all recursive segments ) , we found that recursive introns are spliced more slowly than non-recursive introns of similar size ( S5F Fig; Mann-Whitney P < 2 . 2 × 10−16 ) , consistent with the larger number of biochemical steps involved in recursive versus non-recursive splicing of an intron ., To ask whether recursive splicing occurs while the intron is continuing to be transcribed , we calculated the ratio of the half-life of the first segment to the estimated time needed to transcribe the remainder of the intron ( Methods ) ., For 49% of recursively spliced introns , the first segment half-life is shorter than the time to transcribe the full recursive intron ( Fig 4C ) , implying common co-transcriptional splicing in about half of cases ., We observed that longer recursive introns were more likely to be spliced co-transcriptionally ., The accuracy of splicing is likely to be at least as important as its speed , since splicing to an arbitrary ( incorrect ) splice site will most often produce an mRNA that is unstable or encodes a protein that is aberrant or nonfunctional ., As a simple measure of potential splicing errors , we tallied the fraction of nascent RNA reads ( from the 5 minute labeling period ) that spanned “non-canonical” splice junctions , involving pairs of intron terminal dinucleotides other than the three canonical pairs “GT-AG” , “GC-AG” and “AT-AC” that account for ~99 . 9% of all known fly introns ., For the bulk of non-recursive introns ( most of which are < 100 nt in length ) , the frequency of such non-canonical splicing was negligible ( Fig 4D , black curve ) ., However , for non-recursive introns with lengths matching the much more extended lengths of recursively spliced introns , potential splicing errors were much more frequent ( Fig 4D , gray curve ) , suggesting that the fly spliceosome loses accuracy as intron length ( and the number of possible decoy splice sites ) increases ., Notably , recursive introns had ~37% fewer non-canonical junctions compared to similarly sized non-recursive introns ( Fig 4D , gold curve , Kolmogorov-Smirnov P = 0 . 015 ) ., Therefore , presence of recursive splice sites may increase the accuracy of splicing , perhaps at the expense of splicing speed ., Analysis of intermediates can provide insight into otherwise hidden biochemical pathways ., Here , application of new computational approaches to nascent RNA sequencing data , which is highly enriched for splicing intermediates , enabled us to identify about four times more recursive sites in the Drosophila genome than were known previously ., The surprisingly widespread occurrence of recursive splicing raises questions about what functions it may serve ., A priori , this pathway might improve the speed or accuracy of splicing , or might impact regulation ., Our analyses suggest that recursive splicing does not in fact increase splicing rates , and may actually slow splicing somewhat , likely because of the additional steps involved ., However , we observe that the Drosophila splicing machinery appears to make a relatively high rate of errors in the splicing of longer introns , and that presence of recursive sites may substantially improve splicing accuracy ., In splicing of a non-recursive 30 kbp intron , the 5 splice site is synthesized about 20 minutes before its correct partner 3 splice site , creating a long window during which splicing can only occur to incorrect 3 splice sites , likely contributing to the higher error rate seen for long fly introns ., Presence of a recursive site may help to organize the processing of the intron , keeping the splicing machinery associated with the 5 splice site engaged in a productive direction and avoiding engagement with decoy 3 splice sites ., It was previously observed that masking a recursive splice site in a zebrafish cadm2 intron does not change the overall splicing of the intron but reduces cadm2 mRNA levels 5 ., This observation could be explained if the recursive site promotes accurate splicing and prevents unproductive splicing pathways that result in unstable products targeted by RNA decay pathways such as nonsense-mediated mRNA decay ., Recursive sites may also participate in splicing regulation ., A previous study of a handful of recursively spliced introns in humans identified RS-exons that are initially recognized during recursive splicing via an “exon definition” model of splice site recognition 5 , while an alternative “intron definition” pathway has been proposed for recursive splice site recognition in flies 4 ., An exon definition model would require presence of a 5 splice site downstream of each recursive site ., Consistent with this model , we observed that recursive segments following recursive sites are enriched for strong 5 splice site motifs relative to first recursive segments and relative to non-recursive introns matched for length ( Fig 4E ) ., Use of an exon definition pathway in the initial steps of spliceosome assembly might also contribute to splicing accuracy , with the downstream 5 splice site helping to specify the recursive site 9 ., It could also produce alternative mRNA isoforms containing an additional exon 5 ., Exon definition of recursive segments through transient RS-exons requires that the recursive site first be recognized as a 3 splice site and subsequently as a 5 splice site for splicing of the subsequent segment ( assuming that simultaneous recognition of an RS in both modes is sterically prohibited ) ., For this ordered recognition to occur ( and for sequential splicing of recursive sites generally ) , binding of dU2AF/U2 snRNP must outcompete binding of U1 to the RS prior to its splicing to the upstream exon ., Consistent with this expectation , the 3 splice site motifs of RS are very strong , stronger than non-recursive 3 splice sites , and they have higher information content than RS 5 splice sites ( S6 Fig ) ., Developmental genes are enriched for long introns , which are more likely to be recursively spliced , but explanations for this pattern remain murky ., It is possible that intron length is used to tune the timing of expression of these genes relative to the rapid embryonic cell cycle 12 , 13 ., Alternatively , long introns may be needed to accommodate large transcriptional enhancers or complex three-dimensional organization of these gene loci related to their dynamic transcriptional regulation , or to facilitate alternative splicing ., Thus , it is unclear whether recursive splicing is a feature of developmental genes or exists to facilitate the splicing of long introns that independently persist in developmental genes ., In addition to producing unstable mRNAs , splicing errors may also produce stable mRNAs that encode aberrant protein forms , including dominant negative forms ., Perhaps recursive splicing has been selected for in these genes to improve splicing accuracy and avoid production of aberrant developmental regulatory proteins at critical stages to improve the robustness of development ., We used RNA-seq data from our recent study of splicing kinetics in Drosophila S2 cells ( GEO GSE93763; 9 ) ., These data included 3 independent replicates of S2 cells labeled for 5 , 10 and 20 minutes with 500 μM 4-thiouridine , isolation of labeled RNA , and library preparation using random hexamer priming following ribosomal RNA subtraction ., Separation of total RNA into newly transcribed and untagged pre-existing RNA was performed as previously described 14 , 15 , where 4sU-labeled RNA was selectively biotinylated and captured using streptavidin coated magnetic beads ., cDNA for two independent biological replicates of “total” RNA were prepared using an equal mix of random hexamers and oligo-dT primers from unlabeled S2 cells 9 ., Libraries were sequenced with paired-end 51 nt reads ( 100 nt reads for the “total” RNA samples ) , generating an average of 126M read pairs per library ., Reads were filtered and mapped to the Drosophila melanogaster dm3 reference assembly as described in 9 ., Gene expression values ( TPMs ) in each replicate library were calculated using Kallisto 16 and the transcriptome annotations from FlyBase Drosophila melanogaster Release 5 . 57 17 ., We used three features of recursive sites found in our nascent sequencing data to identify recursive sites: ( 1 ) splice junction reads derived from putative recursive sites ( “RachetJunctions” ) , ( 2 ) recursive-site spanning pairs , specifically read pairs that map to sites flanking putative recursive segments such that the fragment length can only be accounted for by the presence recursive intermediate ( “Rachet Pair” ) , and ( 3 ) a sawtooth pattern in intronic read density ( “RachetScan” ) ., Details of the computational and statistical methods for each of these approaches and our pipeline for recursive site detection are described below ., Out of the full set of recursive sites that were identified across all three methods , we filtered down to a final set of sites with the following criteria: ( 1 ) in genes with TPM ≥ 1 in the total RNA libraries , ( 2 ) in introns with at least 3 reads spanning the 5 to 3 splice sites ( using the largest annotated intron ) , and ( 3 ) not overlapping with an annotated 5 splice site in the that intron ., We ran our final pipeline on reads pooled across replicates and labeling periods to increase detection power ., This resulted in a total of 539 recursive sites identified by any method ., High-confidence sites were identified by the criteria used by Duff et al . 4 ., We wrote a script to plot the read density around putative recursive sites and manually filtered each site based on the presence of a recognizable sawtooth pattern ., This resulted in the identification of 243 high-confidence sites ., Conservation of recursive sites was estimated using per nucleotide phastCons scores 18 from a 15-way Drosophila alignment downloaded from UCSC Genome Browser ., We calculated position weight matrices ( PWM ) for the intronic portions of Drosophila 5 and 3 splice sites using all annotated splice sites ., These weight matrices were then juxtaposed with the 3 splice site PWM followed by the 5 splice site PWM to create a recursive splice site motif PWM ., Individual motif occurrences were scored using a normalized bit score 23 ., The bit score for each motif occurrence is defined as the sum across the log probabilities for each nt being drawn from the motif ., We calculated normalized scores by subtracting the minimum possible score and dividing by the range of possible bit scores ., We searched for reads crossing the 5SS-branch point junction using code previously developed in our lab ( https://github . com/jpaggi/findbps ) ., In short , our approach works by: ( 1 ) identifying reads that do not have a valid alignment; ( 2 ) splitting the unalignable reads just before the 7-mer best matching the consensus 5SS motif; and ( 3 ) mapping this split read as a pair , requiring that the second segment align upstream of the first segment ., We require the following features to be present: ( 1 ) each segment must be at least 15 nt long; ( 2 ) only 1 mismatch is allowed per segment; ( 3 ) segments must be separated by less than 1 Mbp; and ( 4 ) the pair has a unique alignment ., We then filtered the resulting alignments for cases where the second segment aligned immediately downstream of a 5SS or recursive site and the first segment aligned within 100 nt upstream of a 3SS or recursive site ., Overall , we detected 323 5SS-branch point junction reads across 319 introns ., The putative branch points show a motif favoring an A at the branchpoint and a U at the -2 position , consistent with the human branchpoint consensus motif ., We observed 7 5SS-branchpoint junction reads from introns that we report to be recursively spliced ., These counts are consistent with analysis by Duff et al . , which identified 46 recursive lariat junction reads amongst 10 . 2 billion reads ., If such reads occurred at the same frequency in our data , we would expect to observe 1 . 8 recursive lariat junction reads ., All implicated branch points are adenosines ., These 7 reads implicate a lariat associated with the following categories of splicing events ( S2 Table ) : ( 1 ) 5SS-RS: two reads associated with two unique junctions; ( 2 ) RS-RS: one read associated with one junction; and ( 3 ) 5SS-3SS: four reads associated with three unique junctions ., The 5SS-3SS lariat junction reads suggest that recursive splicing is not always used for these introns ., All three such junctions derived from introns of lengths far shorter than typical recursive introns ( 1762 nt , 1929 nt , and 2548 nt ) , suggesting that non-recursive splicing may compete with recursive splicing of introns in this size range ., Nascent RNA was isolated after 5 minutes of labeling with 4sU ( as described above ) and reverse transcribed to first-strand cDNA using ProtoScript II Reverse Transcriptase ( M0368S , NEB ) primed with random hexamers according to manufacturer’s protocol ., cDNA was diluted 1:5 and 1uL was used as template for PCR reactions using primers designed to amplify recursive segments anchored by either the intronic 5 splice site or intronic 3 splice site ( S3 Table ) ., PCR amplification was performed using Taq DNA Polymerase ( 10342020 , Invitrogen ) for 40 cycles ., PCR products were visualized on a 1 . 5% agarose gel relative to Azura PureView 50bp DNA ladder ( AZ1155 , Azura ) ., PCR products were purified with a DNA Clean & Concen | Introduction, Results, Discussion, Methods | Recursive splicing , a process by which a single intron is removed from pre-mRNA transcripts in multiple distinct segments , has been observed in a small subset of Drosophila melanogaster introns ., However , detection of recursive splicing requires observation of splicing intermediates that are inherently unstable , making it difficult to study ., Here we developed new computational approaches to identify recursively spliced introns and applied them , in combination with existing methods , to nascent RNA sequencing data from Drosophila S2 cells ., These approaches identified hundreds of novel sites of recursive splicing , expanding the catalog of recursively spliced fly introns by 4-fold ., A subset of recursive sites were validated by RT-PCR and sequencing ., Recursive sites occur in most very long ( > 40 kb ) fly introns , including many genes involved in morphogenesis and development , and tend to occur near the midpoints of introns ., Suggesting a possible function for recursive splicing , we observe that fly introns with recursive sites are spliced more accurately than comparably sized non-recursive introns . | The splicing of RNA transcripts is an essential step in the production of mature mRNA molecules , involving removal of intron sequences and joining of flanking exon sequences ., Introns are usually removed as a single unit in a two-step catalytic reaction ., However , a small subset of introns in flies are removed via splicing of multiple distinct consecutive segments in a process known as recursive splicing ., This pathway was thought to be quite rare since intermediates of recursive splicing are seldom detected ., In this study , we developed three new computational approaches to identify sequence reads , read pairs and patterns of read accumulation indicative of recursive splicing in Drosophila melanogaster cells using data from sequencing of nascent RNA captured within minutes after transcription ., We used these methods to identify hundreds of previously unknown sites of recursive splicing , occurring commonly in fly introns longer than 40kb and often in genes involved in morphogenesis and development ., We observed that recursive splicing is associated with increased splicing accuracy of long introns , which are otherwise often spliced inaccurately , potentially explaining its widespread occurrence in long fly introns . | sequencing techniques, invertebrates, animals, invertebrate genomics, animal models, drosophila melanogaster, model organisms, experimental organism systems, genome analysis, sequence motif analysis, molecular biology techniques, rna sequencing, drosophila, research and analysis methods, sequence analysis, genome complexity, bioinformatics, gene expression, rna splicing, molecular biology, insects, gene ontologies, animal genomics, arthropoda, biochemistry, rna, eukaryota, rna processing, nucleic acids, database and informatics methods, genetics, biology and life sciences, genomics, computational biology, introns, organisms | null |
journal.pgen.1005189 | 2,015 | Extreme Recombination Frequencies Shape Genome Variation and Evolution in the Honeybee, Apis mellifera | In most sexual eukaryotes , average recombination rates do not greatly exceed one crossover per chromosome arm , which is commonly a minimum requirement for correct meiosis 1 ., However , the honeybee , Apis mellifera , has extremely high recombination rates , averaging 19–37 cM/Mb 2–4 , which corresponds to more than 5 crossovers per chromosome pair per meiosis ., Such high rates are observed in other social insects but not their solitary cousins 5 , 6 ., This suggests that high recombination rates are an adaptation favoured by eusociality although the specific causes are unknown ., Insight into this question can be gained by analysing the fine-scale landscape of recombination rate variation in order to understand the molecular mechanisms that govern it ., The molecular mechanisms that determine the genomic distribution of recombination events in honeybees are unclear ., In a wide range of species , recombination events are strongly clustered into short hotspots a few kb in length 7–10 ., In human and mouse , these hotspots are found to be enriched for a DNA motif recognised by the protein PRDM9 11–14 ., This protein binds to the DNA motif and catalyses a histone modification that acts as a mark for the formation of a DNA double stranded break in the same location 15 ., In species without an active PRDM9 , hotspots are often present , but other features may define them ., For example , in dog , where PRDM9 is inactive , recombination events are clustered in un-methylated CpG islands 16–18 ., In yeast and Arabidopsis recombination hotspots are observed in nucleosome-depleted open chromatin and gene promoters 8 , 9 ., The few invertebrate genomes analysed so far tend to lack extreme recombination hotspots 19 , 20 ., In particular , recombination rates in the fruit fly Drosophila melanogaster appear to be less variable across the genome than other species where fine-scale genetic maps are available 20–23 ., Genetic maps of the honeybee do not indicate the presence of hotspots with extremely elevated rates 2 , 3 , 24 or the presence of enriched sequence motifs 4 ., This is consistent with the absence of a PRDM9-like mechanism controlling recombination rates in insects and suggests that other factors are more important ., One such factor could be DNA methylation ., Unlike fruit flies , the honeybee has an intact methylation system 25 , 26 ., It is therefore possible that rates of recombination in the honeybee genome are influenced by DNA methylation patterns , as observed in some other taxa 16–18 , 27 ., In a diverse range of species , local rates of crossing-over correlate with genetic diversity but not with genetic divergence 28 , 29 ., These correlations are inferred to be due to an indirect effect of recombination due to the interaction between selection and linkage and their strength can be used to make inferences about the pervasiveness of natural selection ., Positive selection on favourable mutations or negative selection against deleterious changes reduce levels of linked variation by the processes of genetic hitchhiking and background selection and these effects are predicted to be larger in regions of low recombination 30 , 31 , resulting in lower genetic diversity in these regions ., Strong correlations exist in many species of fruit fly that have been used to predict that large proportions of the genome are affected by selection 32–34 , whereas in humans such correlations are weaker 35 , 36 , suggesting a less pervasive impact of selection on genetic variation ., Social insects such as honeybees have lower effective population sizes than solitary ones 37 , 38 and it is unclear if selection has a similarly pervasive impact on genome variation ., A number of hypotheses have been proposed to explain the extremely high recombination rates in honeybees and other social insects ., One class of hypotheses suggests that they represent an adaptation important for the evolution of behavioural phenotypes in the worker caste ., This could be because the evolution of eusociality entailed rapid evolution and specialisation of workers 39 ., Alternatively , high intra-colony variability in worker phenotypes could be beneficial because it results in a more efficient workforce 40 , 41 ., These factors could lead to increased recombination rates in the vicinity of genes specifically involved in worker phenotypes ., Some studies have reported evidence for higher recombination rates in genes with worker-biased expression 4 , 39 ., However , the cause of these associations is unclear and several questions remain ., In particular , it is not known whether worker-biased genes are preferentially located in regions of high recombination , or whether there is a direct influence of gene expression or a related process on recombination rate within genes ., Recombination can have profound affects of genome evolution via GC-biased gene conversion ( gBGC; reviewed in 42 ) ., This process is believed to occur due to the biased repair of nucleotide mismatches that occur in heteroduplex DNA generated from pairing of two alleles during meiotic recombination ., This involves a small bias towards repairing a mismatch involving a G/C ( or S , for strong ) nucleotide paired with an A/T ( or W , for weak ) nucleotide in favour of retaining the S allele , which results in an increased probability of transmitting the S allele into the gametes ., There is a large amount of indirect evidence that this process occurs , indicating that genomic regions of high recombination accumulate GC-biased nucleotide substitutions over evolutionary time 43 , 44 , which results in a correlation between recombination and GC content 45 ., A transmission bias towards S alleles has also been directly observed in yeast 46 by analysis of the products of meiosis and humans 47 by analysing transmission through pedigrees ., The population dynamics of gBGC are equivalent to selection acting to increase the fixation probability of weak-to-strong ( WS ) mutations 48 ., As such , gBGC can have effects on the site frequency spectrum 49–51 and rate of nucleotide substitution 52–54 similar to selection ., It can also interfere with the process of natural selection ., For example , gBGC could cause increased substitution rates in functional regions that can be mistaken for positive selection 52–54 ., It can also lead to fixation of deleterious changes , including those underlying genetic disease in humans 55 , 56 ., The transmission bias caused by gBGC also results in a skewed allele frequency spectrum , where WS mutations segregate at higher frequencies ., Glémin et al . 57 modelled this property to estimate the strength of gBGC in the human population ., The average strength of gBGC , B , was estimated as 0 . 38 , ( where B = 4NEb , NE being the effective population size and b the gBGC coefficient ) but 1% to 2% of the genome was estimated to be subject to strong gBGC with B>5 ., Phylogenetic estimates indicate variation in B over two orders of magnitude among placental mammals 58 ., The extreme recombination rates of honeybees could also indicate that gBGC is also very powerful , suggesting it could significantly impact molecular evolution in honeybees ., In support of this previous studies found elevated frequencies of WS mutations , particularly in regions of high GC content 2 , 39 ., A striking and unique feature of the honeybee genome is the over-representation of CpG dinucleotides 26 ., The statistic CpGO/E measures the frequency of CpG dinucleotides in a nucleotide sequence compared to its expected value based on individual frequencies of Cs and Gs ., In species where most CpG sites in the genome are methylated , as is the case in plants and vertebrates , CpG sites occur at a much lower frequency than expected due to the effects of methylated CpG hypermutability ( average CpGO/E in humans is 0 . 2 ) and this value rarely exceeds one in eukaryote genomes ., The honeybee genome is unique in that it has a much higher frequency of CpG sites than expected ( CpGO/E is around 1 . 67 ) ., The reason for this is unclear , but possible explanations are a mutational bias in favour of CpG sites or a fixation bias due to gBGC that favours the fixation of mutations that generate CpG dinucleotides ., There are a number of unresolved questions regarding the evolution , molecular control and consequences of recombination in the honeybee genome ., Firstly , is there evidence for recombination hotspots ?, How does gene expression and DNA methylation affect local rates of recombination ?, The answer to these questions could give us insight into how recombination is controlled in invertebrates ., Secondly , does recombination modulate strength of natural selection across the genome ?, This can be addressed by investigating the correlation between recombination rate and the levels of genetic diversity and divergence ., Thirdly , is there evidence for a local increase in recombination rate in the vicinity of genes with worker-biased expression ?, It has been suggested that this could be selectively advantageous due to the importance of worker phenotypes in the evolution of eusociality ., Finally , what effects do the extremely high levels of recombination in honeybee have on the strength of gBGC ?, How does gBGC impact genome variation and the frequency of CpG sites in the genome ?, We can address these questions by analysing the shape of the site frequency spectrum for different SNP categories and estimating the value of B . Here we construct a fine-scale map of recombination rate variation honeybee using population-scale resequencing dataset 37 with the aim of addressing these questions ., Our estimates show good correspondence with a previous genetic map 3 ., Compared to the human genome , recombination events do not appear to be strongly partitioned into hotspots in the honeybee genome ., Our data is consistent with an effect of germline methylation generating variation in crossover rate by suppressing recombination ., We find evidence for a strong association between recombination and levels of genetic variation ., In contrast to previous studies , we do not find that worker-biased expression is a strong predictor of high recombination rate compared to other factors ., We also uncover a major effect of recombination on genome variation via the process of gBGC , which is stronger than observed in any other species and has a major impact on genome variation and evolution ., We constructed a high-resolution map of rates of crossing-over from patterns of linkage disequilibrium among 6 . 2 million SNPs observed across 60 copies of the sixteen nuclear honeybee chromosomes ., We chose to use samples from Africa , sequenced as part of a larger study , because they are the most genetically diverse and because there is no evidence for population structure between them 37 ., We used the LDhat method , which estimates the population-scaled recombination rate ρ across the genome ., This is related to the recombination frequency r by the equation ρ = 3NEr in the case of haplodiploid species , where NE is effective population size ., The LD map contained 306 , 764 discrete rate intervals ., About 50% of the genome is covered by intervals of 5 kb or longer in this map ., The mean recombination rate is 390 ρ/kb and the average rate change is ~9% between adjacent intervals ., Scaling by a NE of 500 , 000 , estimated previously using the same set of African samples 37 , this corresponds to an average crossover frequency , r , of 26 . 0 cM/Mb , which is in agreement with previous estimates of 19–37 cM/Mb 2–4 ., Recombination does not appear to be strongly restricted to a limited portion of the genome ( Fig 1A ) , suggesting that there are not strong hotspots in honeybee genome but a relatively continual recombination landscape ., For example 50% of the recombination events in the genome occur in 32% of the genome ., In humans , a similar map from population scale sequencing suggests that 50% occurs in less than 10% of sequence 59 ., There is however considerable large-scale variation in recombination rates along the chromosomes ( Fig 1B shows variation along Group 1 ) ., The mean population-scaled recombination rate computed from 100 kb windows is 385 ρ/kb , with a standard deviation of 167 ρ/kb ( see S1 Fig for LD maps of all chromosomes ) ., We find that the LD map is broadly congruent with a previously constructed genetic map 3 ( e . g . R2 = 0 . 341 for the large metacentric chromosome Group1; Fig 1B and 1C ) , but the strength of correlations varies among regions and chromosomes ( R2 = 0 . 213 across the genome; Figs 1D and S1 ) ., There is a highly significant correlation between levels of neutral genetic diversity , measured by Watterson’s theta , θw , estimated using noncoding sites , and rates of crossing over in the honeybee genome ( R2 = 0 . 615 , Fig 2A ) ., We also examined the relationship between crossing over rates and divergence between A . mellifera and A . cerana and found a significant but very weak correlation ( R2 = 0 . 018 , Fig 2B ) ., The strong correlation between recombination and genetic variation remains after correcting for divergence ( R2 = 0 . 617 , Fig 2C ) ., These correlations are also found separately in intronic , intergenic and coding regions ( S2 Fig ) ., A highly significant but weaker correlation between diversity/divergence is found using average pairwise heterozygosity ( π ) to measure genetic diversity instead of θw ( R2 = 0 . 480 ) ., A correlation between genetic variation corrected for divergence and recombination rate is consistent with the pervasive influence of linked selection on patterns of variation , due to background selection , recurrent selective sweeps , or both 28 ., It is also possible that fixation biases due to GC-biased gene conversion ( gBGC ) could contribute to the correlation between genetic diversity and crossover rate , as the strength of gBGC is expected to covary with recombination 42 ., To examine this possibility , we recomputed diversity ( using θw ) while removing large classes of non-coding A . mellifera SNPs and substitutions between A . mellifera and A . cerana that may putatively be affected by gBGC ., We first removed all variants that change GC content and found diversity to still be positively correlated with crossover rate ( R2 = 0 . 563 ) ., After observing gBGC among transitions in particular ( see below ) , we next removed all transitions and also observed a positive correlation between diversity and recombination ( R2 = 0 . 586 ) ., These patterns favour linked selection as a major force in shaping variation in the genome ., These correlations are only slightly weaker than the correlations observed when the dataset is randomly subsampled to the same size ( R2 = 0 . 573 and R2 = 0 . 594 respectively ) , which suggests that gBGC has at most a small effect on determining the magnitude of genetic variation in a genomic region ., Average Tajima’s D is negative ( -1 . 178 ) reflecting of skew towards rare variants , as already observed in this African honeybee population 37 , which is indicative of population expansion ., Tajima’s D ( measured in 100 kb windows ) shows a weak negative correlation with both GC content ( R2 = 0 . 114 ) and recombination rate ( R2 = 0 . 015 ) , which indicates a slightly higher skew towards rare variants in regions of high recombination ., Pervasive linked selection is expected to generate a skew towards rare variants in regions of low recombination 32 , which we do not observe ., This could indicate the action of additional factors ., In order to assess whether the association between genetic diversity and inferred recombination rates could be an artefact of having more power to detect recombination in regions of high genetic variation , or due to other biases , we estimated LD-based maps of recombination using datasets where SNPs were removed or using different parameter as follows:, i ) we produced a dataset where genetic variation ( θw ) in each 100 kb window was capped at 0 . 002 , effectively subsampling data in 98% of 100 kb windows;, ii ) we produced a dataset where rare variants ( minor allele frequency<0 . 1 ) were removed;, iii ) we evaluated the effect of increase the block penalty to 10 , which affects the probability of changes in recombination rate between genomic regions ., In each case , the resulting LD maps were strongly correlated with the original map ( S3 Fig ) ., The strong correlation between levels of genetic variation ( in the original dataset ) and inferred rates of crossing over remained in the LD maps produced using different parameters ( S3 Fig ) and are similar to that observed in the original dataset ., We therefore conclude that variation in genetic variation across the honeybee genome does not generate biases in inference of recombination and that the correlation between recombination and genetic variation is real ., Levels of genetic variation are reduced close to genes , indicative of an effect of linked selection 37 ., In order to determine the effects of linked selection acting on coding sequence on our observed correlation between genetic variation and recombination , we analysed this correlation restricting the analysis to sites at different distances from genes ., We find that the correlation between diversity and recombination in intergenic regions is weaker when restricted to sites far from coding sequences ., At sites <20 kb from coding sequences R2 is 0 . 364 , whereas it is 0 . 281 at 50–60 kb and only 0 . 208 at 100–110 kb away ( randomly sampling the same amount of data in each case ) ., This supports the interpretation that this correlation is due to the effect of linked selection , as sites under selection are expected to be rarer far from genes ., This is also supported by a finding of an excess of SNPs with high FST in within coding sequences 37 ., The honeybee genome has low GC content ( on average 34% ) but GC content is variable across the genome 26 ., To further understand the basis for this variation , and how it relates to recombination rate variation , we first partitioned the genome according to annotations and calculated the average GC content among genes and gene elements ., We find that coding , intergenic and intronic regions each have characteristic GC content ( S4 Fig ) ., Coding regions are biased toward high GC content ( 39% ) , whereas intronic regions are particularly low ( 23% ) and intergenic regions are intermediate ( 31% ) ., Interestingly , 5’ UTRs have higher on average GC content than 3’ UTRs ( 31% cf . 24% ) ., A unique feature of the honeybee genome is an overall excess of CpG dinucleotides ( measured by CpGO/E ) , compared to expectations based on frequencies of single bases ., CpGO/E is also highly variable between different functional regions of the genome ., It is high in noncoding regions ( ~1 . 7 in both introns and intergenic regions ) ., However , notably , the coding part of the genome has an average CpGO/E close to the expected ( 1 . 04 ) ., This is consistent with the observation that methylation in honeybees occurs predominantly in gene bodies 60 ., As reported previously 61 , CpGO/E is bimodally distributed among genes ( S4 Fig ) ., We assigned genes into high or low CpGO/E categories compared to the mean of 1 . 19 ., Genes with low average CpGO/E ( LCpG; <1 . 19 CpGO/E ) have high levels of germline methylation at CpG sites and tend to be associated with cellular housekeeping functions , whereas genes in the higher average CpGO/E class ( HCpG; >1 . 19 CpGO/E ) have low levels of germline methylation and tend to be caste and tissue specific 37 , 60–62 ., We find a strong correlation between crossover rates and GC content in the honeybee genome ( R2 = 0 . 436 ) ., Strong correlations are also observed between GC content and crossover rates within coding ( R2 = 0 . 506 ) , intronic ( R2 = 0 . 463 ) and intergenic regions ( R2 = 0 . 446; Fig 3A ) ., These correlations are also observed in 5’ and 3’ UTRs ( S5 Fig ) ., Such , correlations between GC content and crossover rates are observed in a wide variety of taxa , and could suggest that recombination drives GC content via the process of gBGC 42 ., We find that CpGO/E is correlated with recombination in coding sequence ( R2 = 0 . 369 ) but only very weakly correlated in intronic ( R2 = 0 . 066 ) and intergenic regions ( R2 = 0 . 011; Fig 3B ) ., Methylation is mainly restricted to coding sequence in honeybees and variation in CpGO/E in coding sequence is likely to reflect differences in germline methylation 60–62 ., Conversely , variation in CpGO/E in other parts of the genome is not influenced by DNA methylation and also does not correlate with recombination rate ., These results may therefore suggest a role of germline DNA methylation in attenuating recombination rates in the honeybee genome ., Interestingly , we also detect this correlation in 3’ UTRs ( R2 = 0 . 289 ) , but not in 5’ UTRs , ( R2 = 0 . 025; S5 Fig ) which could indicate an effect of differential levels of methylation ., In support of this , the CpGO/E distribution of 3’ UTRs is shifted towards lower CpGO/E compared to noncoding regions , indicative of higher levels of DNA methylation ( S4 Fig; 61 ) ., Average rates of crossing over are reduced in coding sequence and UTRs compared to noncoding regions ( Fig 4A ) ., This suggests the presence of specific factors that reduce recombination specifically within genes ., We next examined how variation in patterns of gene expression and inferred levels of germline methylation are associated with crossover rate in genes ( Fig 4B ) ., Previous studies have suggested that genes with worker-biased expression tend to have high recombination rates 4 , 39 ., To test this , we first compared rates of crossing over within genes with biased expression in queens compared to workers and vice versa 63 ., There were no significant differences in crossover rates between these gene categories ( p = 0 . 61 , bootstrap test ) , although the caste-biased genes had higher than average crossover rates ( 18% increase; p<0 . 01; average for coding regions = 240 ρ/kb ) ., We next compared crossover rates in genes with biased expression in drones compared to workers and vice versa 64 ., Here we found highly elevated recombination rates in worker-biased genes ( 50% increase compared to average; p<0 . 01 ) and decreased recombination rates in drone-biased genes ( 28% decrease; p<0 . 01 ) and unbiased genes ( 23% decrease; p<0 . 01 ) ., We conclude that worker-biased genes have higher recombination rates compared to drone-biased genes , but not compared to queen-biased genes ., These results suggest that genes with elevated expression in both female castes ( queens and workers ) tend to have higher recombination rates , rather than those specifically expressed in workers ., We used two measures to estimate the potential association between levels of germline methylation and rates of crossing over in genes: 1 ) levels of CpGO/E in coding sequences and 2 ) estimates based on direct detection of methylated CpG sites in sperm and egg using bisulphite sequencing 62 ., Genes were classified as HCpG and LCpG based high or low values of CpGO/E as described earlier ., These two measures are highly correlated ., We detect significant methylation in 39% of the LCpG genes and 14% of the HCpG genes ., Out of all genes where we detect methylation , 60% of the CpGs are methylated in the coding sequence of LCpG genes compared to only 18% in the HCpG genes ( S6 Fig ) ., We classified genes as HMET , LMET or UNMET based on the observation of high , low or undetected levels of methylation in the germline ., The HMET category had significantly lower average CpGO/E compared to other categories ., However , the UNMET class has a bimodal distribution of CpGO/E , where 33% of genes have values of CpGO/E <0 . 7 , which could potentially represent germline-methylated genes that were not detected experimentally ., The average crossover rate among LCpG genes is only 29% of the rate esimated in HCpG genes ( p<0 . 01 ) , consistent with an effect of germline methylation suppressing recombination , particularly in HCpG genes ( Fig 4B ) ., Inferred levels of methylation are strongly correlated with patterns of gene expression: female-biased genes tend to be HCpG and highly recombining , whereas male-biased genes tend to be LCpG and have lower recombination rates ., These patterns also correlate with levels of genetic variation: LCpG genes have on average 45% lower genetic diversity than HCpG genes 37 ., The association between levels of recombination and experimentally inferred levels of germline methylation is consistent with these results ., Highly methylated genes have low levels of crossing over , similar to those observed in LCpG genes ., A potential concern is that estimates of ρ made by LDHAT are affected by local variation in NE across the genome , which could lead to underestimation of recombination rate in regions of low genetic variation ., Since ρ and θ are correlated , we conducted additional high resolution scans to test whether the differences in ρ we observe in coding relative to intergenic regions and in LCpG genes relative to HCpG genes could be due biases in inference caused by differences in local genetic diversity between these regions 37 ., We measured ρ and θ in 1 kb windows across the genome ., We found that ρ is consistently higher outside of genes than inside of genes at given levels of θ ( S7 Fig ) ., Likewise , HCpG coding sequences are typically associated with higher ρ than LCpG coding sequences at given levels of θ ( S7 Fig ) , although the difference is less clear in regions where diversity is very high ., We conclude that our inference of ρ in these regions detect significantly different crossover rates that are not merely mirroring local levels of genetic diversity ., We next aimed to test whether the associations between caste biased gene expression , CpG levels , and crossover rates were indicative of specific gene categories being preferentially located in regions of certain recombination rates , or whether the association was restricted to recombination in coding sequences ., Such a regional effect would be predicted if there was a selective advantage for worker-biased genes to occur in regions of high recombination 39 ., We therefore compared patterns of gene expression and methylation to crossover rates in gene-flanking sequence , using 50 kb regions located 10–60 kb from each side of the genes ( Fig 4C ) and in 100 kb regions located 50–150 kb from each side of the genes ( Fig 4D ) ., As expected , crossover rates increase with increasing distance from the gene ( average rate at 10–60 kb distance = 360 ρ/kb; average rate at 50–150 kb distance = 399 ρ/kb ) ., In addition the associations between expression patterns and CpG levels are greatly reduced ., The average decrease in crossover rates of drone-biased genes in flanking regions at 10 kb distance is only 13% compared to all genes and 4% of average >50 kb away ., Crossover rates in the queen vs . worker comparisons are indistinguishable from the average rates in both >10 kb and >50 kb ., The differences in crossover rates between LCpG and HCpG genes are also reduced in flanking regions compared to crossover rates within coding sequence ., There is an 3 . 37x difference in crossover rates between LCpG and HCpG within coding sequence ( p<0 . 01 ) but this is reduced to 1 . 24x and 1 . 07x in the >10 kb and >50 kb flanking regions respectively ., Associations between methylation classes and crossover rate are also significantly weakened in flanking sequence ., These results indicate that crossover rates vary greatly between genes and correlate with both patterns of gene expression and levels of germline methylation ., However , the finding that these associations are restricted to crossover rates in coding regions is indicative of a direct effect of these factors rather than an accumulation of certain types of genes in regions of high or low recombination , which would be predicted if there was an evolutionary advantage of worker genes being located in regions of high recombination 39 ., We tested whether the associations between crossover rates and gene expression and CpG content were independent of each other ., Genes that are biased in workers compared to drones are enriched in the HCpG class , so it is not clear which of these two factors is driving the association with high recombination rates ., We therefore subdivided both datasets of caste-biased genes according to HCpG and LCpG classes ., We found that the large differences in crossover rates in HCpG and LCpG remain irrespective of patterns of gene expression ( S8 Fig ) : for the same gene expression class , crossover rates are 2 . 3–3 . 4x higher in HCpG compared to LCpG genes ., However , within each CpG class , the difference in crossover rates between drone and worker biased genes is smaller ( 1 . 3x higher in HCpG genes and 1 . 8x higher in LCpG genes ) ., Hence , variation in CpG content is the strongest predictor of recombination rate in our dataset ., One interpretation for this finding is that variation in levels of germline methylation is the strongest factor determining variation in recombination rates within genes in the honeybee genome ., However , the associations we observe with crossover rates and gene expression patterns cannot be completely explained as an effect of differences in inferred levels of germline methylation ., The site frequency spectrum in our dataset is dominated by low frequency AT alleles , which make up 80% of the rare variants ( allele frequency<10% ) across all SNPs , but only 51% of common variants ( allele frequency 40–50% ) , a highly significant difference ( p<10-5 , Fisher’s exact test; S9 Fig ) ., By comparing homologous genomic regions between A . mellifera and A . cerana , we were able to infer the probabilities that either allele represented the ancestral or derived state at 2 , 983 , 700 SNPs using a weighted parsimony method ( see Methods ) ., We categorised each allele at a SNP as weak ( A or T ) or strong ( G or C ) ., At strong-to-weak ( SW ) SNPs , the S allele is ancestral and the W allele is derived , whereas weak-to-strong ( WS ) SNPs are defined as the reverse ., The derived allele frequency spectrum consists mostly of strong-to-weak ( SW ) mutations ( 2 , 037 , 148 SNPs ) , and these are strongly biased towards occurring at low frequencies ( Fig 5A ) ., Weak-to-strong ( WS ) mutations are fewer overall ( 719 , 365 SNPs ) , but are shifted toward high frequency or nearly fixed ., Analysis of the proportions of variants of each type across the allele frequency spectrum therefore reveals a decline in SW and increase in WS variants with increasing allele frequency ., WS variants make up 15% of variants at allele frequencies <0 . 1 but 79% of variants at allele frequencies >0 . 9 ( Fig 5B ) ., This highly skewed site frequency spectrum is indicative of a strongly AT-biased pattern of mutation coupled with a fixation bias towards WS mutations ., Such a fixation bias could be generated by a strong effect of GC-biased gene conversion ( gBGC ) , which manifests as a bias towards transmission of GC alleles ., In order to further investigate this process , we quantified the average allele frequencies of a variety of classes of variants ., We found that WS transitions segregate on average at 3 . 6x higher allele frequency in the population than SW transitions ( p<0 . 01; Fig 5C ) ., However , the average frequencies of WS and SW transversions were similar to each other ( 14 . 1% and 14 . 9% , respectively ) and close to the average derived allele frequency in the sample ( 16 . 8% ) ., These results are consistent with a fixation bias driven by WS transitions ( A→G or T→C ) , which could indicate that gBGC specifically targets transitions in the honeybee genome ., A potential mechanism for this could be that heteroduplex mismatches between two alleles formed by a transition are repaired with a greater GC-bias than | Introduction, Results, Discussion, Methods | Meiotic recombination is a fundamental cellular process , with important consequences for evolution and genome integrity ., However , we know little about how recombination rates vary across the genomes of most species and the molecular and evolutionary determinants of this variation ., The honeybee , Apis mellifera , has extremely high rates of meiotic recombination , although the evolutionary causes and consequences of this are unclear ., Here we use patterns of linkage disequilibrium in whole genome resequencing data from 30 diploid honeybees to construct a fine-scale map of rates of crossing over in the genome ., We find that , in contrast to vertebrate genomes , the recombination landscape is not strongly punctate ., Crossover rates strongly correlate with levels of genetic variation , but not divergence , which indicates a pervasive impact of selection on the genome ., Germ-line methylated genes have reduced crossover rate , which could indicate a role of methylation in suppressing recombination ., Controlling for the effects of methylation , we do not infer a strong association between gene expression patterns and recombination ., The site frequency spectrum is strongly skewed from neutral expectations in honeybees: rare variants are dominated by AT-biased mutations , whereas GC-biased mutations are found at higher frequencies , indicative of a major influence of GC-biased gene conversion ( gBGC ) , which we infer to generate an allele fixation bias 5 – 50 times the genomic average estimated in humans ., We uncover further evidence that this repair bias specifically affects transitions and favours fixation of CpG sites ., Recombination , via gBGC , therefore appears to have profound consequences on genome evolution in honeybees and interferes with the process of natural selection ., These findings have important implications for our understanding of the forces driving molecular evolution . | Evolution results from changes in allele frequencies in populations ., The main forces that cause such changes are natural selection and random genetic drift ., However , an additional process , GC-biased gene conversion ( gBGC ) , associated with meiotic recombination , affects the probability that alleles are passed from one generation to the next ., The honeybee , Apis mellifera , has extremely high recombination rates—more than 20 times to those observed in humans ., However , the reason for this is unknown and the effects of such high recombination rates on evolution are not well understood ., Here we use patterns of genetic variation in the genomes of 30 honeybees to infer variation in the rate of recombination across the genome ., We find that recombination rates and levels of genetic variation are strongly correlated , which is indicative of a pervasive impact of natural selection on genetic variation ., We also infer a major role of DNA methylation in determining recombination rates in genes ., Patterns of genetic variation appear to be strongly skewed due to the effects of gBGC , suggesting that recombination generates a bias in transmission of alleles during meiosis ., This process seems to be interfering with the efficacy of selection at removing deleterious alleles and favouring beneficial ones ., Recombination therefore has a huge impact on genetic variation and evolution in honeybees and appears to play a dominant role in genome evolution . | null | null |
journal.pgen.1000766 | 2,009 | Fine-Scale Variation and Genetic Determinants of Alternative Splicing across Individuals | Alternative splicing ( AS ) allows for multiple mRNA isoforms to be transcribed from a single gene locus , potentially creating much greater protein diversity from our roughly 25 thousand human genes 1 ., AS is very common in higher order organisms and , especially through the lens of newer , high throughput technologies , such as oligonucleotide arrays and transcriptome sequencing , we are finally realizing the true extent and importance of AS ., New studies based on deep sequencing and micro-arrays with exon junction probes estimate the proportion of genes undergoing AS in humans between 74% and 94% 2–5 Hence , it is crucial to understand the functions and the regulation of AS if we are to arrive at a real grasp of regulation of gene expression and gene networks ., Although most isoform variation is thought to occur between tissues , many differences exist among healthy individuals in a population 6–8 ., It is likely these differences are of genetic origin and contribute to phenotypic diversity and disease susceptibility ., Many Mendelian disorders , such as cystic fibrosis 9 , have been explained by splicing errors caused by genetic mutations 10 ., This shows the importance of finding more genetically driven isoform variations to understand the genetic causes of complex diseases ., Until recently , AS differences were not detectable with commercially available micro-array platforms ., Due to low probe densities , those platforms only aimed at measuring gene-level expression and targeted mainly the 3′ untranslated region ( UTR ) of genes ., The Affymetrix Exon Array , with its nearly 5 . 5 million exon-targeted probes , is one of the recent genomic tools available for profiling of splicing or transcript initiation/termination differences between human tissues or between individuals ., In the prelude to this work , Kwan et al . 6 showed using Exon-Array expression data from lymphoblast cell-lines of HapMap individuals that cis-acting polymorphisms are associated with many gene-level expression differences and isoform ratio differences between individuals of the HapMap CEPH population 11 ., While the initial analysis detected numerous robust differential splicing events that were genetically controlled , the study did not attempt to identify the actual causal polymorphisms and it was not clear whether it had sufficient statistical power and signal to noise ratio to detect more subtle genetic influences on exon inclusion levels ., Given the high validation rate achieved for AS in the first study , we expected to find many more inter-individual splicing differences deeper , within the statistically less significant candidates and also within the speculative content of the microarray , which targets predicted or rarely expressed exonic regions ., Moreover , we aimed to determine whether it was possible to identify the causative polymorphisms – as opposed to extended regulatory haplotypes – responsible for such changes ., Thus , in order to increase our detection threshold , in the analysis presented here we made use of publicly available , prior AS information ., Using a customized heuristic optimized for the detection of AS events and some visual curation , we selected a large sample of candidate genetically regulated AS events ., Subsequently , we validated these events using RT-PCR , and in order to identify likely causative SNPs , we re-sequenced the genomic DNA around the alternatively spliced regions ., Finally , we used in silico predictions and , in selected cases , minigene assays , to verify the causative nature of the detected polymorphisms ., This new analysis on the data demonstrates that the Exon Array can detect much more subtle splicing differences than initially suggested and that the speculative , non-core probesets , which constitute the majority of the array , contain useful data for AS discovery ., This study constitutes the most in-depth analysis of cis-regulatory heritable splicing differences to date , having validated quantitatively the greatest number of cases and found a likely causative SNP is most cases ., Our results provide new insights in the genetic regulation of splicing , demonstrate that subtle differences in alternative splicing between humans are more frequent than initially detected , and that the causative polymorphisms can be identified and validated in reporter mini-gene systems ., To optimize our chances of finding true AS events , we only considered candidate probesets whose target genomic coordinates overlapped an AS event catalogued in at least one database , including KnownAlt 12 , ASAP-II 13 and additional events inferred directly from EST/mRNA genome annotations ( see Methods ) ., We ranked candidates statistically according to both the regression p-value and a custom-designed measure representing the unexpectedness of the probesets fold-change in the context of the other probesets in the gene ( see Methods ) ., Further , by visually inspecting probeset fold-changes and regression P-values of the top candidate genes in the UCSC Genome Browser , we selected 68 new potential AS events , along with 4 events from previous analyses 6 , for further validation and characterization ., For the purpose of this analysis , we chose events that would be easily amplifiable via PCR , therefore excluding large intron retentions or alternative transcript initiation/termination events ., The UCSC Genome Browser Tracks of selected candidates are available in Table S1 ., The red “AS-marker” track indicates the position of the affected exon ., In order to detect isoform ratio differences quantitatively as well as qualitatively , we validated our candidate events using semi-quantitative RT-PCR ., Instead of performing the validations on all 57 individuals , we chose a strategic sub-sample of 10 individuals from the HapMap CEPH population , which was the smallest possible sample for which at least one individual was polymorphic at every SNP associated with an AS event ., The primers were targeted to the two exons flanking each alternatively spliced region ., Product abundance was quantified as described in the Methods ., Figure 1A and 1B show examples of electrophoresis readings from which the isoform ratios were estimated ., Out of the 58 candidates which produced interpretable results , 22 showed a significant association with the SNP ( p<0 . 05 , Figure 1B shows an example ) , confirming that the AS event is under the control of a cis-regulatory mutation , 10 showed a visible trend in the expected direction , 7 showed clear differences in isoform ratios between individuals but with no obvious trend linked to the SNP genotype , and 3 showed visible isoforms with no detectable differences in ratios ., The remaining 16 showed no evidence of AS ., It should be noted that the small sample size ( 10 individuals ) and limited representation of different genotypes limits the power of this validation approach ., Thus , while observing a single PCR product in all samples can be considered as a reliable indication of a false positive result , observing two alternatively spliced products of the expected sizes , even without achieving statistical significance , is evidence supporting the initial microarray finding ., Thus depending on the stringency of the validation criteria – detecting a statistically significant association within the PCR data , versus observing two alternatively spliced products - our validation rate is between 38% and 72% ., Table 1 shows all validated AS events organized by the strength of the validation evidence ., For all the candidates selected in this study and a few additional validated candidates from the previous analysis 6 , we sequenced a region of 600–800 bps around the putative AS events in 2 individuals predicted to preferentially express one isoform and 2 individuals expressing the other ., The selection of individuals was based on both the genotype of the associated SNP and the micro-array expression scores for the probeset of interest ., Thus , even if the associated SNP was not perfectly linked to the causative SNP , selecting for extreme expression phenotypes increased the chances that the 4 individuals would differ at the causative polymorphism ., Analysis of the sequencing results revealed 86 polymorphisms in 60 confidently sequenced regions , 76 of which were already catalogued in dbSNP129 14 , and 10 which were novel ., Out of the 34 validated AS events which were successfully sequenced ( including 4 from the previous analysis ) , all showed at least one SNP within the sequenced interval , as did 6 out of the 7 sequenced regions which were negatively validated ., For each gene in which one or more SNPs were identified , Table S2 links to custom UCSC Genome Browser tracks which indicate the position of SNPs and the associated fold-change in sequenced individuals ., For 24 of the genes , we could identify a SNP within 20 bases of a relevant splice-site ., Using the MaxEntScan algorithm 15 , which calculates the theoretical strength of splice-sites based on maximum entropy , we could verify whether the expected effects of many SNPs adequately explained the difference in micro-array expression between the sequenced individuals ., In all 17 cases where differences in splice site strength could be calculated , the micro-array results and the maximum entropy scores of the polymorphic splice-site sequences agreed on the direction of the effect ( See Table 2 ) ., Figure 2 shows the types of AS events and the relative position of the affected splice-site , which is crucial to understanding the direction of the expression changes ., For an additional 16 cases where a SNP was present within the affected exon , we used the ESE Finder 3 . 0 online tool 16 to predict ESEs affected by the SNP and assign them a matrix-based affinity score ( see Table 3 ) ., In 13 out of the 16 exons , the predicted change in the number and/or affinity of ESE motifs was concordant with the probeset expression change ., Although , in some of these cases , identifying the affected splice-site is not obvious , we are still able in all cases to infer the direction of the predicted probeset expression change , as shown in Figure 2 ., The 3 for which the predicted effect was in disagreement were the 3 cases with the smallest expression fold-change ., This result is encouraging , particularly since it only concerns the binding preferences of 4 splicing factors , out of possibly dozens , and the less than perfect agreement between predictions and results likely reflects the relative lack of detailed understanding of ESE motifs as compared to splice-site consensus sequences ., Out of the 4 validated and sequenced AS regions remaining , 1 had a SNP inside the retained intron ( HNRPH1 ) and 3 had one or more intronic SNPs around the cassette exon ( IL6 , WDR67 and SIDT1 ) ., It is likely that intronic SNPs play a role in determining isoform levels through their effect on intronic splicing enhancers ( ISEs ) or silencers ( ISSs ) ., We used another online software tool , SpliceAid 17 , to detect ISEs or ISSs but only one of the 3 cases , showed qualitative agreement with the expression data ( data not shown ) , indicating that we are either looking at the wrong candidate SNPs or that our understanding of ISEs/ISSs is not yet detailed enough to predict the effect of all these polymorphisms ., Since we only sequenced approximately 200 base pairs from each exons , and introns generally span thousands of bases , causative SNPs are likely to be found further away in the intron ., Intronic SNPs were found in many cases in conjunction with exonic SNPs , but there were often multiple intronic SNPs ., Since we only looked for qualitative agreement between in silico predictions and the observed splicing differences , looking at many SNPs per gene would surely have caused more chance correlations and we would have been unable to rank the effects of SNPs affecting different splicing factors ., For this reason , we prioritized exonic or splice-site bordering SNPs , for which there was almost always a single candidate ., In order to assess whether the candidate SNPs truly cause the splicing differences in vivo , for 6 genes we sub-cloned the exon and surrounding intronic sequence from individuals of different genotypes into a minigene expression vector 18 ., Placing the exon of interest between two constitutive exons within the reporter construct allows determining the effect of the SNP on candidate exon inclusion levels ., We used sequencing to verify that the sub-cloned construct differed only at the SNP positions we had previously predicted to be responsible for the differential splicing event ., In 5 out of the 6 cases , the putative causative SNP was the only SNP present ., MMAB however contains 3 closely neighboring SNPs which are perfectly linked ., We transiently transfected the constructs into HeLa cells and assessed the presence of different isoforms using RT-PCR ., Figure 3 shows the electrophoresis band migrations along with the SNP genotypes ., For CAST , ERAP2 , and PARP2 , for which the candidate SNP is very close to the splice-site , we demonstrate that the predicted SNP causes a complete switch in 5′ splice-site usage ( PARP2 , ERAP2 , Figure 3A and Figure 2B ) or a complete skipping of the exon ( CAST , Figure 3C ) ., In the 3 other cases , ATP5SL , MMAB and AMACR , for which the candidate SNPs were found in the exon and were predicted to disrupt ESEs , the assay shows a visible but subtle change in isoform ratios ( Figure 3D ) in the expected direction ., We used the Agilent 1000 DNA chip to quantify the results for these three cases because their gel bands are less convincing than the first three cases ., We included the capillary electrophoresis readings as Figure S1 ., The quantification of the isoforms from the peaks show that the smallest difference between individuals of different genotypes , the first and last columns for the AMACR gene in Figure 3 , consists of a 1 . 5 fold difference in isoform ratios , confirming that all the changes in isoform ratios were measurable and in the expected direction ., Why significant variation exists between individuals with the same genotype is hard to say , especially considering that the plasmid inserts were confirmed to have the same sequence ., The differences must come either from changes incurred during transfection ( a single transfection assay was performed for each plasmid ) or from biological or technical noise ., Except for MMAB , for which any one of the 3 SNPs or even a combination of all 3 could be causative , these results demonstrate the causative nature of our candidate SNPs and confirm that both systematic splicing differences as well as more subtle , quantitative , differences exist between individuals and the extent of the change is reflective of the position of the causative SNP and of in silico predictions of its effect ., In order to gain insight into the persistence within human populations of SNPs with such drastic effects on the structure of the expressed transcripts , we wanted to assess whether some of the causative SNPs showed signs of recent positive selection ., We performed two tests ., Using the program fdist2 19 , we calculated the fixation index ( Fst ) of all 41 re-sequenced HapMap SNPs based on the frequencies in the 4 available populations ( see Methods ) , and compared the results to a simulated dataset ., The average Fst was significantly greater than expected ( p\u200a=\u200a0 . 046 ) by a small margin ., One SNP had a Fst above the 99 . 999th percentile relative to the simulation dataset , rs6580942 ( A/C ) , the best candidate SNP in the ESPL1 gene ., This is very significant , even in a sample of 41 ( p\u200a=\u200a4 . 1E-3 ) ., The C allele has a frequency of 0% in the African ( YRI ) and Asian ( CHB+JPT ) populations compared to 33% in the Caucasian population ( CEU ) ., We then calculated the highest relative extended haplotype homozygosity ( REHH ) in the CEU population considering various size blocks below 500 kb centered on the same 41 SNPs ( see Methods for details ) ., We performed the same analysis on 300 randomly chosen SNPs in genes as a control ., There was no significant difference between the average highest REHH of our sequenced SNPs and the control set ., No SNP showed a REHH score above the 99th percentile of the control set ., However , we noticed that rs10941112 ( A/G ) , the best candidate SNP from the AMACR gene , had both a Fst above the 97th percentile , with the A genotype varying from 0% in the African population to 37% and 62% in the Asian and Caucasian populations respectively , and a REHH above the 97% percentile , with the A allele showing 122 . 5 fold greater homozygosity for a block of 391 kB around the SNP than the G allele ., The combined extreme result in both tests is highly unlikely ( p\u200a=\u200a8 . 0E-4 ) and is indicative of positive selection ., These two analyses have brought to light strong evidence suggesting that the SNPs rs6580942 and rs10941112 , in the ESPL1 and AMACR genes , respectively , have undergone recent positive selection ., Other than these 2 SNPs , there is no strong evidence of positive selection , indicating that many of our re-sequenced HapMap SNPs may be selectively neutral ., Previous micro-array studies have often attempted to estimate the real amount of AS or gene-level differences simply from counting the number of cases which surpass a multiple-testing corrected significance threshold 6 , 8 , placing complete faith in the results of the micro-array as well as the normalization , summarization and whichever AS detection algorithm was used ., The first problem with such an approach is that many sources of noise can cause false positives , like SNPs within probes , cross-hybridizations or technical noise , as well as other distortions such as probesets responding unevenly to a gene-level change ., The second weakness of such estimations , as our results demonstrate , is that many real AS events lie far below standard significance thresholds ., The highest P-value for a validated AS candidate in this study was 7 . 3E-4 as opposed to 4 . 2E-9 in the previous study on the same dataset ., This means that approximately 15 times more probesets could be considered potential AS candidates ., Of course , we do not suggest that we found 15 times more AS than the previous study but rather , by integrating EST evidence and utilizing a more sophisticated AS detection algorithm , we can show that the speculative content of the array , which comprises about 80% of the array , as well as the less significant measurable differences on the array contain valuable information which have thus far , been mostly overlooked ., Our results show that Exon-Array data by itself may be too noisy to produce reliable estimates of AS at the genome-wide scale ., All we can conclude is that genetically controlled splicing differences exist between individuals and are probably more common than was previously estimated ., A more definitive answer on the real extent of individual-specific AS may come from high-throughput sequencing , which can avoid probe target bias and detect AS differences qualitatively rather than through statistical inference ., We have shown that many splicing differences between healthy individuals can be identified using the Exon Array platform ., We expect that many more , perhaps approaching the complete map of the splicing eQTL ( expression quantitative trait loci ) landscape , will be catalogued soon using more sensitive methods , such as deep mRNA sequencing ., Most of these splicing differences which we can detect are controlled by polymorphisms in cis-regulatory regions or in the vicinity of an implicated splice-site ., These differences between individuals could contribute to phenotypic variation and could either be neutral in their effects , or confer differential susceptibility to complex diseases ., A few of our validated events occur in genes which have already been associated with diseases ., BCKDHA is related to maple syrup urine disease , type 1a 20 , a rare inherited metabolic disorder which , without a highly controlled diet and close monitoring of blood chemistry , causes progressive neurological damage which can cause vomiting , eating difficulties , irregular breathing , coma or death ., Deficiency of the gene alpha-methylacyl-coa racemase ( AMACR ) is a rare disorder of the fatty acid metabolism which is characterized by neuronal and liver abnormalities 21 and the gene is considered a useful biomarker for various types of cancer 22 , making it quite interesting that this gene contained the SNP with the most evidence of positive selection ., Interleukin 6 ( IL6 ) is an important mediator of fever 23 and the gene has been associated with osteoporosis 24 and Kaposis sarcoma 25 ., MMAB is related to vitamin B12 responsive methylmalonic aciduria 26 , the inability to synthesize adenosylcobalamin , a vitamin B12 derivative , and whose symptoms include metabolic acidosis and retarded development ., A SNP in MMAB , which is in linkage disequilibrium with our causative splicing SNP , has recently been associated with HDL cholesterol levels 27 ., Surprisingly , all of the above AS events are within protein-coding regions of the genes , making it very likely that these heritable differences contribute to individuals predisposition to disorders similar to those caused by inactivation of those genes or to other , more complex , diseases ., Complex diseases such as diabetes , cancer or schizophrenia are expected to be influenced by polymorphisms in a large number of genes , which may interact in multifarious ways ., Many of the polymorphisms we identified in this study induce , through AS , potentially much more dramatic changes to the protein sequence than do non-synonymous coding SNPs ., We have shown that common SNPs can influence alternative splicing across individuals even in relatively important genes , indicating that the genotyping of such SNPs will likely play a very significant role in predicting the occurrence of complex diseases in the future ., Hundreds of genome-wide association studies ( GWAS ) carried out to date have generally failed to identify causative protein-coding disease variants 28 ., Many of the underlying causes may be due to subtler , regulatory genetic influences 29 ., Thus , a lot of interest and resources have been allocated to identifying eQTLs , genes whose expression levels are affected by regulatory SNPs ., Although identifying eQTLs has been quite successful 6 , 30–33 , there has been considerably less accomplishments in teasing apart regulatory haplotypes and pinpointing the SNP actually responsible for the regulatory difference ., Our groups earlier work demonstrated the existence of common isoform eQTLs; i . e . genes under genetic control resulting in differential expression of transcript isoforms including: alternative splicing , alternative polyadenylation , and alternative transcript initiation ., Here , we postulate that in many of those cases , it should be possible to narrow down the regulatory region to the vicinity of the alternative event ( e . g . cassette exon ) , and subsequently identify the causative polymorphism with high confidence ., In some of the cases , as in the ERAP2 gene , the common splicing polymorphism introduces a premature stop codon , most likely resulting in nonsense-mediated decay of the alternative product and a drastic reduction in the overall transcript levels , suggesting that splicing and RNA processing variations may be underlying some of the common expression QTLs ., This knowledge will be essential for future functional studies and perhaps for future applications such as genetic therapy ., In 2004 Nembaware et al . used public EST data to show the existence of allele-specific transcript isoforms in human 34 ., In 2007 , Hull et al . showed that it is possible to identify such events in lymphoblastoid cell lines 7 ., They selected 70 alternative splicing events , and showed that 6 of them were consistently associated with a specific genotype ., They also used in vivo assays demonstrating causative nature of 2 candidate SNPs , suggesting that a substantial number of alternative splicing events may be controlled by genetic polymorphisms ., These studies were followed by genome-wide microarray-based approaches 6 , 8 , 35 , which confirmed and further expanded our knowledge of genetic control of isoform variation ., Earlier this year , Zhang et al . published an article 8 where they used the Exon-Array on lymphoblast cell lines to look for genetic variants which account for AS differences between populations ., They claim , based on multiple-testing-corrected statistics , that they discovered 397 such differences between the Caucasian ( CEPH ) and African ( YRI ) populations ., Recently , other groups have approached the problem from the purely genetics angle: knowing the polymorphisms that are present , they attempted to predict their effect computationally and identify exons that are differentially spliced across individuals ., This approach has so far met with limited success ., Elsharawy at al ., 36 have obtained an extremely low validation rate of their in silico predictions , ranging from 0% for ESE predictions to 9% for SNPs in splice-sites , demonstrating our far from complete understanding of the effects of cis-regulatory sequences on splicing ., Thus , in the present study , we take further steps towards optimal integrated use of the existing data – gene structure annotation , splicing-sensitive microarray data , SNP databases and targeted sequencing - to detect splicing eQTLs and their genetic determinants ., First , taking advantage of the high level of coverage of the current sequence-based annotation of AS events , we concentrate only on events that have been previously reported ., This approach is highly justified by the observation from the previous analysis 6 that less than 10% of the detected and validated AS events were novel ( unannotated ) , suggesting that the current EST coverage of the transcriptome is nearly complete ., Secondly , we use a much improved algorithm to detect AS events in exon array data ., Finally , we show that among the events that are detected using the above criteria and further validated in the lab , a majority contain SNPs that have highly suggestive in silico evidence of causation ., We can show in a post hoc analysis that knowing the sequence variation information before-hand could have significantly improved the specificity of our search for AS ., 27 out of 34 ( 79% ) validated and sequenced alternatively spliced regions contained a SNP for which in silico evidence appropriately explained the change , compared to 2 out of 7 ( 29% ) for the sequenced regions which were negatively validated ( data not shown ) ., Given the current low resolution of HapMap SNPs , making use of in silico predictions of SNPs effects at the genome-wide level would only be applicable to a fraction of Exon Array probesets ., Less than half of our likely causative SNPs ( in splice-sites or exons ) were HapMap SNPs ., The experience of Elsharawy et al . showed that the specificity of the purely computational approach is quite low , given the current level of understanding of AS regulation ., However , once the resolution of SNPs and their genotypes increases significantly , which is already the case for four CEPH HapMap individuals which were recently fully sequenced ( http://www . 1000genomes . org/ ) , it should become feasible to merge computational predictions with biological expression data to improve the power to detect cis-acting polymorphisms involved in splicing ., In turn , the identified polymorphisms and their effects will help to further enhance our understanding of splice-sites and cis-regulatory sequences ., Out of the two splice-sites , the 5′ intronic splice-site ( the donor site ) appears to be the dominant identifiable target of these regulatory polymorphisms , with 14 predicted causative SNPs , as opposed to 3 for the 3′ splice-site ( the acceptor site ) ., Given that we considered 23 bases for the 3′ splice-site compared to 9 , for the 5′ splice-site , it seems unlikely the result of pure chance , suggesting either that there may be greater purifying selective pressure acting on the region around the 3′ splice-site , or that , due to the greater degeneracy of the 3′ sequence 15 , SNPs affecting it influence splicing strength too subtly to be detected by the micro-array platform ., In the case of cassette exons , the fact that the 5′ splice-site sequence , as well as ESE sequences , influence the use of the upstream splice-site defining the exon start , demonstrates the fact that in multi-intronic genes in vertabrates , the “exon definition” step , whereby splice-sites are paired across the exon , takes place before the assembly of the mature spliceosome and splicing of the intron can occur 37 , 38 , as opposed to species like yeast , whereby the intron definition occurs independently for each intron and a mutation of the 5′ splice-site would cause retention of the downstream intron rather than exon skipping 39 ., The fact that the vast majority of putative causative SNPs affected either the 5′ splice-site or ESE sequences shows that the exon definition plays a central role in generating these splicing variations across individuals ., Expression QTL analysis has garnered considerable interest in recent years and is increasingly being used in conjunction with whole genome association studies to narrow down the list of genetic variants putatively responsible for complex genetic disorders 28 ., Here , we focus on a specific type of eQTL , alternative splicing variation , and extend the results of prior studies by validating the greatest number of these differences and showing that such variation may be more common than previously estimated , and that its effects can be quantitatively very subtle ., Furthermore , this work demonstrates that we have the ability to identify the specific causative genetic variants responsible for isoform eQTL differences among individuals ., It also underscores the value of data integration in order to obtain improved true positive rates in large scale analyses of splicing ., With the upcoming release of the 1000 genomes data 40 and the growing use of high-throughput sequencing for transcriptome analysis , we will be able to broaden our understanding of the complex intricacies of the “splicing code” and use this knowledge to confidently identify splicing regulatory SNPs ., In order to select alternative splicing candidates within the vast set of significant associations between SNPs and probeset expression , we used publicly available knowledge of AS events to prioritize the list ., We only considered Exon-Array probesets whose target coordinates overlapped an EST-supported AS event ., We used a cut-off of a single base , reasoning that a single base mismatch could significantly affect the probe binding efficiency 41 ., We downloaded the list of putative AS events in Human based on UCSC Known genes and human AS events from ASAP-II ., We also retrieved from the UCSC Genome Browser website the full tables describing the human genome annotations based on Blated 42 spliced EST and mRNA sequences to gather some additional AS evidence ., We applied strict criteria when inferring potential AS events from EST/mRNA data ., To avoid confusing intron retention events with incomplete splicing or transcript length changes with incomplete mRNA sequences , we only selected cassette exons or alternative splice site usages ., The latter two types of AS are also most confidently validated via PCR since they generally introduce short changes in the mRNA sequence ., We had to define a reasonable gene structure from the Blat results , which , especially for EST annotations , include many short gaps which most probably originated from sequencing errors rather than genuine splicing ., We considered gaps greater than 30 bps as introns , ignored gaps smaller or equal to 3 bases , and filtered out annotations containing gaps o | Introduction, Results, Discussion, Methods | Recently , thanks to the increasing throughput of new technologies , we have begun to explore the full extent of alternative pre–mRNA splicing ( AS ) in the human transcriptome ., This is unveiling a vast layer of complexity in isoform-level expression differences between individuals ., We used previously published splicing sensitive microarray data from lymphoblastoid cell lines to conduct an in-depth analysis on splicing efficiency of known and predicted exons ., By combining publicly available AS annotation with a novel algorithm designed to search for AS , we show that many real AS events can be detected within the usually unexploited , speculative majority of the array and at significance levels much below standard multiple-testing thresholds , demonstrating that the extent of cis-regulated differential splicing between individuals is potentially far greater than previously reported ., Specifically , many genes show subtle but significant genetically controlled differences in splice-site usage ., PCR validation shows that 42 out of 58 ( 72% ) candidate gene regions undergo detectable AS , amounting to the largest scale validation of isoform eQTLs to date ., Targeted sequencing revealed a likely causative SNP in most validated cases ., In all 17 incidences where a SNP affected a splice-site region , in silico splice-site strength modeling correctly predicted the direction of the micro-array and PCR results ., In 13 other cases , we identified likely causative SNPs disrupting predicted splicing enhancers ., Using Fst and REHH analysis , we uncovered significant evidence that 2 putative causative SNPs have undergone recent positive selection ., We verified the effect of five SNPs using in vivo minigene assays ., This study shows that splicing differences between individuals , including quantitative differences in isoform ratios , are frequent in human populations and that causative SNPs can be identified using in silico predictions ., Several cases affected disease-relevant genes and it is likely some of these differences are involved in phenotypic diversity and susceptibility to complex diseases . | Alternative splicing ( AS ) , through the alternative use of exons , can produce many different mRNA transcripts from the same genomic locus , thus possibly resulting in the production of many different proteins ., We know that splicing differences between individuals exist and that these changes are often associated with genetic variants ., Thus far , very few of these associations have led to the precise localization of the causative polymorphisms ., In this work , using in-depth analysis of previously published splicing sensitive micro-array data from human cell lines , we identified and validated a large number of splicing changes which are highly correlated with nearby genetic variations ., We then sequenced the genomic DNA around candidate exons and used in silico modeling tools to identify causative SNPs for most of our candidates ., Using a plasmid reporter construct , we further demonstrated that five selected SNPs reproduce the expected effect in vivo ., Our results indicate that genetically controlled splicing differences between individuals may be more common than previously suggested and can be very subtle; and most are caused by SNPs affecting either the splice-site region or exonic splicing enhancers ( ESEs ) sequences . | molecular biology/rna-protein interactions, computational biology/molecular genetics, computational biology/alternative splicing, molecular biology/rna splicing, molecular biology/bioinformatics, genetics and genomics/genetics of disease, computational biology/genomics | null |
journal.pgen.1003660 | 2,013 | Global Analysis of the Sporulation Pathway of Clostridium difficile | Clostridium difficile is a Gram-positive , spore-forming , obligate anaerobe that causes gastrointestinal diseases including diarrhea , pseudomembranous colitis , and toxic megacolon 1–3 ., C . difficile infections and C . difficile-related deaths have risen dramatically in the past decade , increasing the financial burden on health care systems 4–7 ., While C . difficile is best known for causing hospital-acquired antibiotic-associated infections , recent epidemiologic studies indicate that community-acquired C . difficile infections are increasingly more common and associated with significant morbidity 6 , 7 ., A key element to the success of C . difficile as a pathogen is its ability to produce spores ., Spores are resistant to most disinfectants and antibiotics , making them difficult to eliminate both from infected humans and the environment 1 , 2 , 8 ., As a result , C . difficile spores disseminate readily from person to person and cause high rates of recurrent infections , which can lead to serious illness or even death 1–3 , 9 ., Although spores are critical to the pathogenesis of C . difficile , their composition and formation remain poorly characterized ., Less than 25% of the spore coat proteins identified in the well-characterized spore-former Bacillus subtilis have homologs in C . difficile 10 ., In contrast , the regulatory proteins that control spore coat gene expression and other sporulation events in B . subtilis are conserved in C . difficile and all other spore-forming Firmicutes 10–13 ., These include the master sporulation transcriptional regulator , Spo0A , and the sporulation sigma factors σF , σE , σG , and σK ., In B . subtilis the sporulation sigma factors function at discrete stages during spore development to couple changes in gene expression with specific morphological changes in the cell 14–16 ., The morphological changes begin with the formation of a polar septum , which creates two compartments , the mother cell and the forespore ., The mother cell engulfs the forespore and guides the assembly of the spore until it lyses once spore maturation is complete ., By coupling these developmental changes to the sequential activation of compartment-specific sporulation sigma factors , the mother cell and forespore produce divergent transcriptional profiles that coordinately lead to the formation of a dormant spore 16 ., Sporulation gene transcription in B . subtilis begins with the activation of the transcription factor Spo0A , which in turn activates early sporulation gene transcription , such as the genes encoding the early sigma factors σF and σE ., σF is initially held inactive by an anti-σ factor and only undergoes activation after septum formation is complete; this mode of regulation couples σF activation in the forespore to a morphological event 17 , 18 ., Active σF induces the transcription of genes whose products mediate cleavage of an inhibitory pro-peptide from σE in the mother cell via trans-septum signaling 19 ., Active σE induces the transcription of genes whose products lead to the activation of the late sporulation sigma factor σG in the forespore , which occurs during or after engulfment 20 , 21 ., Activated σG in the forespore subsequently induces the expression of genes whose products proteolytically activate σK in the mother cell via trans-septum signaling 22 ., Notably , the activity of each sigma factor relies on the activation of the preceding sigma factor 11 , 14–16 , 23 ., As a result , the sigma factors operate in a sequential , “criss-cross” manner and collectively control the expression of hundreds of genes during sporulation 24–26 ., The regulatory pathway controlling sporulation sigma factor activation in B . subtilis is thought to be conserved across endospore-forming bacteria , since all four sigma factors are conserved 11 , 12 ., However , a growing body of work in the Clostridia suggests that diverse pathways regulate sporulation sigma factor activity in the Firmicutes ., In C . perfringens , a sigG− mutant still produces cleaved σK , suggesting that σG does not control the proteolytic activation of σK as it does in B . subtilis 27 ., Furthermore , a C . perfringens sigK− mutant exhibits a phenotype more severe than a B . subtilis sigE− mutant in that it fails to initiate asymmetric division or produce σE 28 , suggesting that in C . perfringens σK functions upstream of σE ., Indeed , C . perfringens σE and σK have been suggested to be dependent on each other for full activity , in contrast with B . subtilis 28 ., A similar early sporulation defect has been observed in a sigK− mutant of C . botulinum , which also exhibits reduced expression of early sporulation genes spo0A and sigF 29 ., In contrast with B . subtilis and C . perfringens , however , a C . acetobutylicum sigF− mutant does not initiate asymmetric division 30 , and a sigE− mutant fails to complete asymmetric division 31 ., In addition , a C . acetobutylicum sigE− mutant produces wildtype levels of σG 31 in contrast with B . subtilis , and a sigG− mutant exhibits elongated forespores and pleiotropic defects in coat and cortex formation 31 ., To determine how these sporulation sigma factors regulate sporulation in C . difficile , we constructed mutations in the genes encoding the sporulation transcription factor Spo0A and the sigma factors σF , σE , σG , and σK and determined the transcriptional profiles of these mutants using RNA-Sequencing ( RNA-Seq ) ., The transcriptional analyses , combined with cytological characterization of the sigma factor mutants , suggest that divergent mechanisms regulate the activity of σG and σK in C . difficile relative to B . subtilis and other Clostridium spp ., In addition , these analyses have identified a set of 314 genes that are upregulated during sporulation in a Spo0A- , σF- , σE- , σG- , and/or σK-dependent manner ., These sporulation-induced genes provide a framework for identifying and characterizing C . difficile spore proteins that may have diagnostic or therapeutic utility ., In order to identify genes that are regulated by the sporulation-specific sigma factors , we used a modified TargeTron gene knockout system to disrupt the genes encoding σF , σE , σG , and σK in C . difficile 32 ., This system uses a group II intron to insert an erythromycin resistance cassette into the target gene ( Figure S1A ) ., JIR8094 33 , an erythromycin-sensitive derivative of the sequenced C . difficile strain 630 34 , was used as the parental strain ., As a control , we also constructed a targeted disruption in spo0A , which encodes the master regulator of sporulation Spo0A 35 , 36 ., Colony PCR of the intron-disrupted mutants confirmed the expected size change resulting from the intron insertion into the spo0A , sigF , sigE , sigG , and sigK genes ( Figure S1B ) ., To determine the effect of blocking sigma factor production on sporulation , the mutants were induced to sporulate on solid sporulation media and visualized by phase contrast microscopy 37 ., It should be noted that sporulation is asynchronous in this assay , and the extent and timing of sporulation exhibits variability even between biological replicates ( Figure S2 ) ., Nevertheless , after 18 hrs of growth , sufficient numbers of cells have initiated sporulation to detect the production of immature phase-dark forespores and mature phase-bright spores in the wildtype strain ( Figure 1 and S2 ) ., In contrast , spo0A− , sigF− , sigE− , sigG− , and sigK− cultures failed to produce phase-bright spores ( Figure 1 ) ., No phase-dark or phase-bright forespores were observed in the spo0A− , sigF− , or sigE− strains , suggesting a block early in sporulation ., Analysis of live , sporulating cultures with the lipophilic dye FM4-64 ( to stain mother cell and forespore membranes ) and Hoechst 33342 ( to stain cell nucleoids ) revealed polar septum formation in wild type and the sigma factor mutants but not in the spo0A− mutant ( Figure 1 ) ., This result was consistent with the observation that Spo0A is necessary to induce the sporulation pathway in C . difficile 35 , 36 ., Overall , the proportion of sporulating cells detected by membrane and DNA staining in the culture was 25% , 41% , 24% , 26% , and 18% for wildtype , sigF− , sigE− , sigG− , and sigK− , respectively , as indicated by the presence of a polar septum , immature forespore compartment , or mature forespore ( Table S1 ) ., Wildtype cultures contained a heterogenous population of sporulating cells at discrete stages of sporulation: 28% of sporulating cells exhibited intense DNA staining of an FM4-64-labeled forespore compartment ( yellow arrows , Figure 1 , Table S1 ) ; 28% showed phase-dark forespores that stained with both FM4-64 and Hoechst ( Table S1 ) , 28% exhibited phase-dark forespores that stained intensely with FM4-64 but not Hoechst ( green arrows , Figure 1 , Table S1 ) , and 16% contained a phase-bright forespore that failed to be stained with either FM4-64 or Hoechst ( pink arrows , Figure 1 , Table S1 ) ., In contrast , sigF− and sigE− sporulating cells were arrested at the asymmetric division stage , with 95% and 92% of sporulating cells , respectively , exhibiting intense DNA staining of an FM4-64-labeled forespore compartment ( yellow arrows , Figure 1 , Table S1 ) ., The sigG− mutant strain was arrested at the phase-dark forespore stage , with 69% of sporulating cells exhibiting intense forespore membrane and nucleoid staining ( yellow arrows , Figure 1 , Table S1 ) ., While only 4% of the sigG− cells were observed to produce forespores that stained only with FM4-64 , 44% of sporulating sigK− cells were captured at this stage of sporulation , a phenotype that was also observed in wildtype ( green arrows , Figure 1 , Table S1 ) ., Taken together , these results indicate that all four sporulation sigma factors are required to complete spore formation and suggest that σG is necessary to complete the stage of sporulation development required to exclude the Hoechst dye from staining the forespore chromosome ., The results are also consistent with studies investigating B . subtilis forespore development , which indicate that nucleic acid stains are excluded earlier than membrane stains during spore development 38–40 ., To confirm that the gene disruptions prevented sigma factor production in each of the respective sigma factor mutants , we performed Western blot analyses using antibodies raised against C . difficile sigma factors ., Similar to B . subtilis , Spo0A was required for the production of all the factors , and σF was observed in the sigE− , sigG− , and sigK− strains at wildtype levels ( Figure 2 , 41 ) ., σE was detected in both its pro- and cleaved form in wildtype , sigG− and sigK− strains , whereas the majority of σE was unprocessed in the sigF− strain ( Figure 2 ) ., This result slightly deviates from the B . subtilis model , where pro-σE processing is completely abrogated in a B . subtilis sigF− strain 42 ., In contrast , a C . perfringens sigF− mutant fails to produce pro-σE altogether 27 , and σE processing has not been demonstrated in C . acetobutylicum 31 ., σK was present in wildtype and sigG− mutant strains but absent in the sigF− and sigE− strains ( Figure 2 ) , analogous to observations in B . subtilis where σE is required for sigK expression ., A C . perfringens sigE− strain in contrast produces low amounts of σK 28 ., Consistent with the observation that C . difficile σK lacks an N-terminal pro-peptide 43 , no processing of σK was observed in wildtype C . difficile ( Figure 2 ) , even though σK undergoes proteolytic activation in B . subtilis and C . perfringens 27 ., σG was detected in the C . difficile sigF− , sigE− and sigK− mutants ( Figure, 2 ) in contrast with studies of other endospore-forming bacteria , where σG activity and auto-activation of sigG transcription is partially dependent on σE in B . subtilis 44–46 , and σG production depends on σF in C . perfringens and C . acetobutylicum 27 , 30 ., We next performed transmission electron microscopy ( TEM ) to identify the precise developmental stage at which each sigma factor mutant was stalled ., Cortex and coat layers were present on forespores in wildtype sporulating cells , while the spo0A− mutant exhibited no signs of spore formation ( Figure 3 ) ., The sigF− mutant failed to progress beyond asymmetric division ( Figure 3 ) , similar to a B . subtilis sigF− mutant 47 but in contrast with a C . acetobutylicum sigF− mutant which does not initiate asymmetric division 30 ., Nevertheless , unlike B . subtilis , a more electron-translucent region in the mother cell cytosol surrounded by electron dense layers was observed in some sigF− mutant cells; this region resembled mislocalized spore coat ( 37 , Figure 3 ) ., The C . difficile sigE− strain was arrested at the asymmetric division stage similar to the sigF− mutant , although electron-translucent regions surrounded by coat-like layers were not observed in any sigE− cell analyzed ., The C . difficile sigE− mutant phenotype resembled the phenotype of sigE− mutants of B . subtilis 47 and C . perfringens 28 , with frequent observations of disporic cells or cells with multiple septa at one pole ( Figures 1 and 3 ) ., This observation was in contrast with a C . acetobutylicum sigE− mutant , which does not complete asymmetric division 31 ., The C . difficile sigG− mutant produced forespores lacking an apparent cortex layer , similar to B . subtilis 21 , 44; however , unlike B . subtilis , the forespores were surrounded by thin layers that resembled the spore coat layers visible in wildtype cells ( Figure 3 ) ., In addition , the C . difficile sigG− mutant exhibited pleiotropic defects including forespore ruffling , incomplete membrane fission during engulfment , and a septated forespore compartment ( Figures 3 and S3 ) ., Quantitation of the prevalence of each phenotype revealed that forespore ruffling , incomplete engulfment , and a septated forespore compartment were observed in 98 , 87 and 21% of sigG− cells , respectively ( Figure S3 ) ., Lastly , the C . difficile sigK− mutant produced forespores surrounded by a layer that resembled the cortex layer of wildtype , but no coat layers were apparent ( Figure 3 ) ., This phenotype was more similar to a B . subtilis sigK− mutant , which lacks both cortex and coat 22 , than C . perfringens , which fails to initiate polar septum formation 28 ., To validate that the observed mutant phenotypes were due to the targeted insertions , we complemented the mutant strains by expressing a wildtype copy of the gene encoding the corresponding sigma factor from a plasmid ., We used either the pMTL83151 or pMTL84151 multicopy plasmids 48 to express the complementing genes or operons from their native promoters ., The complementation constructs all restored production of phase-bright spores when expressed in their respective mutant backgrounds ( Figure S4A ) , although phase-bright spore formation by the sigK complementation strain was delayed relative to wildtype ., Western blot analysis further confirmed that the complementation constructs restored production of the respective sigma factor to wildtype levels ( Figure S4B ) ., TEM analysis revealed that all four complementation constructs restored coat and cortex formation to their respective mutant strains ( Figure S5 ) ., Heat resistance assays to measure complementation strain sporulation efficiency revealed that the sigF− and sigE−constructs fully complemented heat resistance relative to wildtype and that the sigG− and sigK− constructs partially complemented heat resistance ( 70 and 23% , respectively , Figure S4C ) ., While these analyses showed that σF , σE , σG , and σK were all required for mature spore formation , they did not reveal which genes were being misregulated in the sporulation sigma factor mutants to produce their respective sporulation defects ., To identify these genes and gain insight into the regulatory network controlling sporulation sigma factor activity , we used RNA-Sequencing ( RNA-Seq ) to transcriptionally profile our sporulation mutants and wild type during sporulation ., Three biological replicates of wildtype , spo0A− , and sporulation sigma factor mutant strains were grown on sporulation media ( Figure S2 ) , and RNA was isolated ., Following DNase-treatment , ribosomal RNA depletion and reverse transcription , Illumina-based RNA-Seq was used to determine the complete transcriptome of wildtype C . difficile and the sporulation mutants ., Genome coverage and sequencing counts for each strain and replicate can be found in Table S2 ., The DeSeq variance analysis package 49 was used to identify genes that were downregulated by ≥4-fold with an adjusted p-value of ≤0 . 05 in the spo0A− strain relative to wild type ., This pair-wise analysis identified 276 genes as being Spo0A-dependent ( Table S3 ) ., Consistent with the role of Spo0A as the master regulator of sporulation , 65 of these genes were predicted to be involved in sporulation ( Table S4 ) 11 , 50–52 ., Six of these Spo0A-dependent genes were recently identified as encoding components of the C . difficile spore coat 50 , 53 , and 36 sporulation-related genes ( Table S4 ) were shown to depend on σH , the stationary phase sigma factor that induces spo0A transcription in C . difficile 51 and B . subtilis 54 ., σF- , σE- , σG- , and σK-dependent genes were identified by comparing the transcriptional profiles of the sigF− , sigE− , sigG− , and sigK− strains to wild type , respectively , using the same parameters as above ., This analysis identified 183 genes as being dependent on σF for their expression ( Table S5 ) ., One hundred eighteen of these σF-dependent genes were also σE-dependent ( Table S6 ) , indicating that σE has some activity in a sigF− mutant consistent with the reduced levels of cleaved σE being detected by Western blot ( Figure 2 ) ; 29 of the σF-dependent genes formed a separate subset of genes that were also σG-dependent but σE-independent ., Indeed , the majority of the 34 σG-dependent genes identified in this analysis were not dependent on σE for their expression ( Table S7 ) , since only four of the σG-regulated genes were also σE-regulated ., Notably , none of the genes identified as being σG-dependent required σK for their expression ( Table S8 ) , suggesting that the σG produced in the sigE− and sigK− mutants is active ( Figure 2 ) ., This result differs from the B . subtilis model where σE is needed to fully activate σG function 20 , 21 , 46 , 55 , 56 ., Of the 169 genes that depended on σE for their expression ( Table S6 ) , 85% and 78% of these genes were dependent on Spo0A and σF , respectively ( Figure 4 ) ., The expression of 29 of these genes was also σK-dependent ( Table S5 ) ., Indeed the majority of the 31 σK-dependent genes were σE-dependent ( Table S8; Figure 4 ) , consistent with σE being required for σK production ( Figure 2 ) ., In contrast , as described earlier , no overlap was observed between σG- and σK-dependent genes ( Figure 4 ) ., Taken together , the RNA-Seq analyses suggested that ( 1 ) a small subset of σF-dependent genes are neither σE , σG , nor σK-dependent; ( 2 ) σE activity depends on Spo0A and σF but not σG or σK; ( 3 ) σK activity depends on Spo0A , σF , and σE but not σG , and ( 4 ) σG activity depends on Spo0A and σF but not σE or σK ., The latter two findings differ from the B . subtilis model , where the σK-dependent genes are also σG-dependent because σK activity depends on σG 11 , 15 , 22 , and σG-dependent genes are σE-dependent because full activation of σG requires σE 20 , 21 , 46 , 55 , 56 ., To visually represent the differences in gene expression profiles between the sigma factor mutants and wild type , we generated a heat map for genes downregulated by ≥4-fold with an adjusted p-value of ≤10−5 in the spo0A− strain relative to wild type ., The expression levels of wild type and the sigma factor mutants relative to spo0A− strain were centered , scaled , and mapped to a red-green color scale ., The heat map revealed a cluster of genes that was poorly expressed in the sigE− mutant relative to the wildtype , sigG− , and sigK− strains; these genes were also expressed at reduced levels in the sigF− mutant ( Figure, 5 ) and were primarily σE-dependent ( Table S4 ) ., A separate cluster of genes was downregulated in both the sigK− and sigE− mutants relative to the wildtype and sigG− strains ( Figure 5 ) ; these genes were all identified as σK-dependent genes ( Table S5 ) ., Another discrete cluster of genes was downregulated in the sigG− and sigF− strains relative to the wildtype , sigE− , and sigK− strains ( Figure 5 ) ; again , most of these genes were identified as σG-dependent genes , although two genes were σF-dependent but not σG-dependent ( Tables S5 and S7 ) ., Thus , identification of variably expressed genes between the strains confirmed the findings of our earlier pair-wise analyses: σF-dependent genes were largely Spo0A-dependent , σE-dependent genes were largely σF-dependent , σK-dependent genes were σE-dependent , and σG-dependent genes were σF-dependent but not σE- or σK-dependent ., These results support a model where ( 1 ) σF controls the activation of both σE and σG , ( 2 ) σE induces the production and activation of σK , and ( 3 ) σE and σK are dispensable for σG activation ., Alternative statistical models were also employed to validate these findings ( see Text S1 and Figures S6 and S7 ) ., To validate the RNA-Seq data , we isolated RNA from three separately prepared biological replicates of wildtype , spo0A− , sigF− , sigE− , sigG− , and sigK− strains grown on sporulation media for 18 hrs ., RNA was reverse transcribed and quantitative RT-PCR ( qRT-PCR ) was performed using primers specific for three genes within each of the sigma factor-dependent transcriptomes ., Gene expression levels in the wildtype and the sigma factor mutant strains relative to spo0A− were determined by comparative CT analysis normalized to the housekeeping gene rpoB ., These analyses confirmed that the transcript levels of the σF-dependent gene gpr was reduced by >50-fold ( p<0 . 0001 ) in the sigF− mutant relative to wild type , and reduced in the sigG− mutant by ∼4 fold ( p<0 . 01 ) ; gpr expression was not affected in the sigE− and sigK− mutants ., cd0125 ( spoIIQ , 13 ) transcription was reduced by >10-fold in the sigF− mutant relative to wild type ( p<0 . 01 ) , but no reduction in transcript levels was observed in sigE− , sigG− , and sigK− mutants ( Figure 6A ) ., Transcription of cd2376 was reduced by 3-fold in the sigF− relative to wild type ( Figure 6A ) ., Although this correlation was not statistically significant , it approached statistical significance ( p\u200a=\u200a0 . 065 ) ( Figure 6A ) ; this result is likely due to the low number of overall cd2376 transcripts present in the samples ., Transcript levels of the σG-dependent genes spoVT , sspB , and dacF showed significant reductions in the sigG− ( p<0 . 0004 , <0 . 0002 and <0 . 0001 , respectively ) and sigF− mutants ( p<0 . 0001 ) compared to wild type but no significant reduction in the sigE− and sigK− mutants relative to wild type ( Figure 6B ) ., This observation was consistent with the RNA-Seq data indicating that σG activity depends on σF , although it is likely that σF directly induces the transcription of some σG-dependent genes given the predicted overlap in their promoter specificities 11 ., Nevertheless , given that σG is present at wildtype levels in a sigF− strain , these observations suggest that σF regulates σG activity through a post-translational mechanism ., σE-dependent genes cd3522 and spoIVA were reduced by >100-fold , and cd1511 by >50-fold , in sigE− relative to wild type , ( p<0 . 0001 , <0 . 0001 , and <0 . 006 , respectively ) , but not in sigG− and sigK− mutants ( Figure 6C ) ., Transcript levels of these σE-dependent genes were reduced by ∼5 to 6-fold ( p<0 . 01 ) in the sigF− mutant relative to wildtype , indicating that , in the absence of σF , σE activity is reduced but detectable ., Transcript levels of the σK-dependent genes cd1433 , cd1067 and sleC were significantly reduced by >100-fold in the sigE− ( p<0 . 0001 for each gene ) and the sigK− ( p<0 . 0001 for each gene ) strains compared to wild type ( Figure 6D ) ., σK-dependent gene expression was reduced in the sigF− mutant by 8 to 10-fold ( p<0 . 01 ) , suggesting that σK has reduced but detectable activity in the sigF− strain ., Importantly , no statistically significant change for any of these σK-dependent genes was observed in the sigG− mutant relative to wild type , consistent with the RNA-Seq results indicating that σK activity does not depend on σG ( Figures 4 and 5 ) ., Altogether , the qRT-PCR data validated the RNA-Seq data identifying σF , σE , σG , and σK-dependent genes and confirmed that ( 1 ) σE , σG , and σK activity depend on σF , ( 2 ) full σG activity requires σF but not σE , and ( 3 ) σK activity requires σE but not σG ., It should be noted however that , although σF is required for full σE and σK activity , some degree of σE- and σK-dependent gene expression is observed even in the absence of σF ., Having validated the RNA-Seq data at the transcript level , we next investigated whether changes in transcript levels correlated with changes in protein levels for σF- , σE- , σG- , and σK-regulated genes ., To this end , we raised antibodies against proteins encoded by genes identified by RNA-Seq as being σF- , σE- , σG- , and σK-dependent ., Western blot analyses of the germination protease Gpr confirmed that only σF is required for gpr expression , while production of the regulatory protein SpoVT and the small acid-soluble protein SspA depended on both σF and σG ., These results indicate that σG can directly activate the expression of spoVT and sspA ( Figure 7 ) ., Western blot analyses for CD3522 , SpoIVA , and CD1511 demonstrated that their production depends on σE but not σG or σK; these proteins were detected , albeit at greatly reduced levels , in the sigF− mutant ( Figure 7 ) ., These results were consistent with the observation that active , processed σE is present in both sigG− and sigK− strains , while only trace amounts of processed σE could be detected in the sigF− strain ( Figure 2 ) ., Analysis of σK-dependent protein production using antibodies specific for CD1433 , CD1067 and SleC confirmed that these proteins were absent in the sigE− and sigK− mutants and present in wild type and the sigG− mutant ( Figure 7 ) ., Only SleC was reliably detected in the sigF− mutant , even though cd1433 and cd1067 transcripts could be detected in the sigF− strain ( Figure 6D ) ., Nevertheless , taken together these observations confirm that ( 1 ) σF does not require σE , σG , or σK for activation , ( 2 ) full σE activation requires σF , ( 3 ) full σG activation requires σF but not σE or σK , and ( 4 ) σK activation requires σF and σE but not σG ., While mutation of all four sporulation sigma factors in C . difficile abrogated functional spore formation as expected 11 , the regulation and function of these sigma factors in C . difficile differed from the regulatory pathways determined for B . subtilis and other Clostridium spp ., The differences between C . difficile , C . perfringens , C . acetobutylicum , and B . subtilis sporulation pathways are summarized in Figure 8 , as are the similarities ., Similar to B . subtilis , our transcriptional and cytological analyses reveal that C . difficile σK functions downstream of σE to regulate late-stage sporulation events , and σG functions downstream of σF to regulate forespore maturation ( Figures 2 and 6 ) ., In contrast with B . subtilis , C . difficile σG is fully active in the absence of σE , and σK is fully active in the absence of σG ( Figures 6 and 7 ) ., The latter observation could have been anticipated given that C . difficile σK lacks an N-terminal pro-peptide , in contrast to all other spore formers 43 ., However , the former observation was unexpected because σE-regulated gene products function to activate σG in the forespore of B . subtilis , initiating a positive feedback pathway that increases σG levels through auto-activation of the sigG promoter 44 , 46 , 57 ., In particular , B . subtilis σG activation requires the formation of a σE-dependent “feeding tube” 20 , 21 , 55 , 58 , 59 , which maintains forespore integrity by transporting small molecules from the mother cell into the forespore 20 , 21 , 55 ., This mode of regulation couples the activation of the forespore-specific σG to σE-controlled events in the mother cell ., In contrast , our results indicate that C . difficile σG is active in the absence of σE-dependent feeding tube gene expression ( Figures 5 and 6 , Tables S6 and S7 ) ., Nevertheless , even though σG was active at wildtype levels in the sigE− mutant ( Figures 6 and 7 ) , it remains possible that σG activity may be mislocalized in the mother cell cytosol , similar to the premature activation of σG in Lon− and anti-σG sigma factor CsfB− cells 57 , 60 , 61 ., Even though C . difficile σG can be fully activated in the absence of σE , our results further show that σG is post-translationally activated in a σF-dependent manner ( Figures 2 and 6 ) ., These results raise the intriguing question as to how σF activates σG independent of σE in C . difficile ., In B . subtilis , multiple post-translational mechanisms control σG activity; however , aside from the feeding tube , these mechanisms are inhibitory rather than activating ., In B . subtilis the Lon protease reduces σG activity in the mother cell 60 , while the anti-σ factors SpoIIAB 57 , 62 and CsfB ( also known as Gin ) 61 , 63 , 64 prevent σG activity in the forespore until engulfment is complete ., Whether these factors inhibit σG activity in C . difficile is unknown , although C . difficile does not appear to encode a CsfB homolog ., In future studies , it will be interesting to determine whether σF functions to activate σG directly or alleviate its inhibition , and whether C . difficile sporulation sigma factors exhibit compartment-specific activity similar to B . subtilis ., Interestingly , the morphology of the C . difficile sigG− mutant differed considerably from a B . subtilis sigG− mutant ., While B . subtilis sigG− mutant forespores are normal in appearance despite lacking both a coat and cortex 44 , C . difficile sigG− mutant forespores produced layers resembling spore coat around the forespore and exhibited defects in engulfment and structural integrity ( Figures 3 and S3 ) ., The forespore membrane ruffling phenotype of C . difficile sigG− mutants was reminiscent of B . subtilis feeding tube mutant phenotypes 21 , suggesting that σG may encode proteins required to “nurture” the C . difficile forespore ., Alternatively , σG could regulate a cytoskeletal or cortex component that confers structural integrity to the forespore ., Such proteins could be represented in the σG-regulated genes identified in this study ( Table S7 ) ., The phenotype of the C . difficile sigF− mutant also differed from its cognate mutant in B . subtilis , since the sigF− mutant produced low levels of σE− and σK− induced gene products ( Figure 7 ) and regions that resembled mislocalized coat in the mother cell cytosol ( Figure 3 ) 47 ., In B . subtilis , σF is required to activate the expression of spoIIR , which encodes an intercellular signaling protein that activates SpoIIGA , the protease responsible for activating pro-σE 65 , 66 ., Whether the trace amounts of σE processing observed in the C . difficile sigF− mutant results from low-level expression of spoIIR or spoIIGA , or whether an unknown protease activates σE , remains to examined ., Comparison of the sporulation pathway of C . perfringens relative to C . difficile indicates that both organisms proteolytically activate σE in a σF-dependent manner ( Figure 2 ) , although it should be noted that a C . perfringens sigF− mutant does not make σE , σG , or σK 27 in contrast with C . difficile ( Figure 2 ) ., Si | Introduction, Results, Discussion, Materials and Methods | The Gram-positive , spore-forming pathogen Clostridium difficile is the leading definable cause of healthcare-associated diarrhea worldwide ., C . difficile infections are difficult to treat because of their frequent recurrence , which can cause life-threatening complications such as pseudomembranous colitis ., The spores of C . difficile are responsible for these high rates of recurrence , since they are the major transmissive form of the organism and resistant to antibiotics and many disinfectants ., Despite the importance of spores to the pathogenesis of C . difficile , little is known about their composition or formation ., Based on studies in Bacillus subtilis and other Clostridium spp ., , the sigma factors σF , σE , σG , and σK are predicted to control the transcription of genes required for sporulation , although their specific functions vary depending on the organism ., In order to determine the roles of σF , σE , σG , and σK in regulating C . difficile sporulation , we generated loss-of-function mutations in genes encoding these sporulation sigma factors and performed RNA-Sequencing to identify specific sigma factor-dependent genes ., This analysis identified 224 genes whose expression was collectively activated by sporulation sigma factors: 183 were σF-dependent , 169 were σE-dependent , 34 were σG-dependent , and 31 were σK-dependent ., In contrast with B . subtilis , C . difficile σE was dispensable for σG activation , σG was dispensable for σK activation , and σF was required for post-translationally activating σG ., Collectively , these results provide the first genome-wide transcriptional analysis of genes induced by specific sporulation sigma factors in the Clostridia and highlight that diverse mechanisms regulate sporulation sigma factor activity in the Firmicutes . | C . difficile is the leading cause of healthcare-associated infectious diarrhea in the United States in large part because of its ability to form spores ., Since spores are resistant to most disinfectants and antibiotics , C . difficile infections frequently recur and are easily spread ., Despite the importance of spores to C . difficile transmission , little is known about how spores are made ., We set out to address this question by generating C . difficile mutants lacking regulatory factors required for sporulation and identifying genes that are regulated by these factors during spore formation using whole-genome RNA-Sequencing ., We determined that the regulatory pathway controlling sporulation in C . difficile differs from related Clostridium species and the non-pathogenic model spore-former Bacillus subtilis and identified 314 genes that are induced during C . difficile spore development ., Collectively , our study provides a framework for identifying C . difficile gene products that are essential for spore formation ., Further characterization of these gene products may lead to the identification of diagnostic biomarkers and the development of new therapeutics . | medicine, developmental biology, infectious diseases, genetics, biology, microbiology | null |
journal.pcbi.1000805 | 2,010 | Rapid Transition towards the Division of Labor via Evolution of Developmental Plasticity | I consider a finite population of asexual haploid cells that form undifferentiated multicellular colonies by binary division ., Mutation occur during cell divisions ., Colonies surviving to the time of reproduction disintegrate; the released cells start new daughter-colonies ., Each cell founding a colony goes through divisions so that the final colony size is cells ., Each cell is characterized by viability and fertility ., The former is a measure of the cells contribution towards the survival of the colony it belongs to , e . g . via flagellar action 20 , 21 ., The latter is defined as the probability that the cell successfully starts a new colony ., I assume the existence of two major genes with effects and controlling cell fertility and viability , respectively ( ) ., The direct effects of these genes increase the corresponding fitness components ., To capture the fundamental trade-offs between cells division and locomotion capabilities 3 , 4 , 22 , I postulate indirect negative effects of on viability and of on fertility ., Specifically , fertility and viability are defined using a simple multiplicative model:In the right-hand side of these equations , the first terms account for the direct effect of genes ., Positive parameter controls the shape of the relationships between direct genetic effect and the corresponding fitness component ., The second terms specify the reduction of a fitness component due to the need to develop/maintain the other trait ., Positive parameter specifies the strength of fitness tradeoffs ( which are completely absent if ) ., Because direct effects of genes are expected to be at least as strong as indirect effects , it is reasonable to assume that ., The population of colonies is subject to density-dependent viability selection; all cells comprising surviving colonies can potentially form their own colonies in the next generation ., Following previous work 4 , 15 , the viability of each colony is defined as the average of viabilities of individual cells ( i . e . ) ., To describe viability selection at the colony level , I use a version of the Beverton-Holt model in which the probability that a colony survives to the time of reproduction depends on its viability and the overall number of colonies in the population:where is the maximum carrying capacity of the population of colonies and parameter gives the number of “offspring” of each colony ., In the deterministic version of the Beverton-Holt model ( which represents a discrete-time analog of the logistic model 23 ) , the population size monotonically approaches the carrying capacity for any positive initial condition ., The probability that a cell from a surviving colony does start a daughter colony is given by its fertility ., By the models assumptions , the carrying capacity of a population of identical colonies isso that increasing cell viability and/or fertility increases the number of colonies and cells maintained in the system; if the colony size is very large , ., Note that in this model there is a conflict between individual level selection which favors larger values of and colony level selection which favors larger values of ., Both and cannot be maximized simultaneously because of the trade-offs ., Mutation occurs during the process of cell division resulting in within- and between colony genetic variation ., I assume each gene mutates with a small probability per cell division ., Note that if a mutation does happen , the expected number of mutant cells per colony is which is approximately 24 ., I assume that mutation changes the corresponding allelic effect ( or ) by a value chosen randomly and independently from a truncated Gaussian distribution with zero mean and a constant standard deviation ( with truncation at and ) ., This is a version of the standard continuum-of-alleles model 25 ., Note that a mutant cell in a colony will benefit if it has a higher value of and/or smaller value of than other cells as this will increase the cells fertility ., However such a cell will decrease the colonys viability ., Next I add a possibility for gene regulation ., Molecular data suggest that in green algae Volvox carteri , which is a bona fide multicellular organism with a complete division of labor between two cell types 26 , the germ-soma differentiation is controlled by three types of genes 20 , 27 , 28 ., First , the gls genes cause asymmetric division resulting in a large number of small cells and a small number of large cells ., Then the regA gene acts in small cells supressing their reproductive development , so that they become soma , and the lag gene acts in large cells supressing their somatic development , so that they become germ ., Note that the expression of the regA gene has been shown to depend on environmental factors 29 ., In the model , I postulate the existence of some dichotomy in the internal and/or external environment of the cells ., For example , it can be asymmetry due to the differences in their size ( large and small ) or in their spatial position ( e . g . inner and outer layer of the colony ) leading to differences in some external stimuli ( e . g . chemical or temperature ) ., I call the two types of cells the proto-germ cells and the proto-soma cells ., I assume that within each colony the proportion of the proto-germ cells is and that of the proto-soma cells is ., I further assume the existence of two differentially expressed regulatory genes with effects and , respectively ( ) ., The first gene ( analogous in action to the lag gene ) , is expressed in the proto-germ cells suppressing the effect of the “viability gene” from to ., The second gene ( analogous in action to the regA gene ) is expressed in the proto-soma cells suppressing the effect of the “fertility gene” from to ., These two genes control the developmentally plastic response of the cell to the gradient in the internal and/or external environment ., Note that in contrast to other modifiers studied in population genetic models 30–32 , the two suppressor genes considered here have direct effect on fitness ., This feature is common in theoretical models of phenotypic plasticity 33–35 ., Since evolving gene suppression mechanisms and developmental plasticity is expected to involve fitness costs 36 , 37 , I assume that fertility of the proto-germ cells and viability of the proto-soma cells are reduced by factors and , respectively ., In numerical simulations I used Gaussian functions:The costs grow as suppression becomes more efficient ( i . e . with deviation of and from zero ) ; positive parameter scales the costs of suppression ( larger values correspond to smaller costs ) ., Gene effects on reproductive and somatic function as well as fertility and viability of the proto-germ and proto-soma cells in the general model are shown in Table 1 ., The initial population of cells have all and values set at so that no gene suppression is present ., I allow for mutation in the regulatory genes and describe its effect in a way analogous to that in the major loci ., The complete germ-soma differentiation corresponds to and all evolving to so that germ cells have maximum fertility but cannot survive on their own while soma cells have maximum viability but cannot reproduce ., First I studied a variant of the general model in which gene regulation was absent ( i . e . , and values were set to zero ) ., I used a multidimensional invasion analysis 38–43 and stochastic individual-based numerical simulations ( see Methods for details ) ., Both methods show that in this model the major gene effects and relatively rapidly evolve towards intermediate values so that both fitness components and the population size are relatively low ( see Figure 1 ) ., The inability to increase fitness is a consequences of fitness trade-offs explicitly accounted for by the model ., Analytical approximations show that the equilibrium values of and satisfy to inequalities ., As the strength of fitness tradeoffs decreases to , both and approach ., As the colony size becomes larger , both equilibrium values converge to ., If , then at equilibrium with given by a solution of an algebraic equation ., In general , analytical and numerical results show that increasing the strength of selection , the strength of trade-offs , and decreasing the colony size result in decreasing both fitness components and the population size ., To analyze the whole model I performed large-scale stochastic individual-based simulations that account for selection , mutation , and random genetic drift ( see Methods ) ., For each run , all individuals in the initial population were genetically identical with the major locus effects and set to values chosen randomly and independently from a uniform distribution on and the suppressor effects and set to zero ., The simulations show that the initial phase of evolution is typically driven by selection on the major loci whose effects evolve towards the optimum values predicted by our theory when developmental plasticity is absent ( as in Figure 1 ) ., After that there are three dynamic possibilities ., First , the population stays at a state in which developmental plasticity is absent ( so that and remain close to 0; Figure 2 , first row ) ., Second , some developmental plasticity evolves but the resulting degree of differentiation between proto-germ and proto-soma cells is intermediate ( Figure 2 , second row ) ., Third , one observes the evolution of strong developmental plasticity and complete germ-soma differentiation ( Figure 2 , third row ) ., The last outcome is observed when costs of developmental plasticity are small , mutation rates are high , and fitness trade-offs are strong ( Figure 3 ) ., The effects of increasing costs of plasticity and mutation rate on the plausibility of differentiation are intuitive ., Indeed , less constraints and more genetic variation typically means more adaptation ., But why do fitness trade-offs have such a big effect ?, This happens because larger values of imply that fitness advantage of a highly differentiated state is larger ., For example , for the parameter values used in the simulations the size of the equilibrium population of undifferentiated colonies is thousand ., However , the size of the equilibrium population of completely differentiated colonies will be about , and thousand for and , respectively ., That is , the benefit of cell differentiation for the population size ( and fitness ) increases dramatically with ., The results shown in Figures 2–3 as well as in Supporting Information ( Text S1 and Figures S1 , S2 , S3 , S4 , S5 , S6 , and S7 ) are for ., If , the conditions for complete differentiation are more strict ., Neither the proportion of the proto-germ cells nor the colony size affect the results qualitatively ., Analytical approximations for the case when the colony size is very large ( i . e . ) allow one to get some additional insights ., In particular , one can find the conditions for stability of a population state with no gene regulation ( i . e . , ) towards introduction of mutations with small positive values of and ., These conditions are illustrated in Figure 4 which shows that this equilibrium becomes unstable so that some gene suppression evolves if parameters and are sufficiently large and the cost of developmental plasticity is low ( i . e . is not too small ) ., Moreover , one can show that if fitness trade-offs are sufficiently strong ( ) then the corresponding dynamic system has an equilibrium in which major effects have maximum possible values ( ) whereas the minor gene effects are ., The later value is biologically feasible ( so that ) , if fitness costs of plasticity are sufficiently high ( ) ., If , only partical gene suppression evolves ., If the costs are relatively low ( ) , the analytical approximations suggest that complete gene suppression evolves ( i . e . , ) ., These results are well in line with numerical simulations described above ., The model introduced and analyzed here shows the emergence of complete germ-soma differentiation ., This is achieved via the evolution of developmental plasticity resulting in the suppression of somatic function in one subset of the colonys cells and of reproductive function in the remaining cells of the colony ., Differential suppression of gene expression is triggered by environmental factors during development ., A necessary condition for this process is the existence of sufficiently strong trade-offs between somatic and reproductive functions significantly reducing fitness ., Also necessary are sufficiently high mutation rates and sufficiently low costs of developmental plasticity ., With parameter values used here , complete germ-soma differentiation can evolve within a million generations ., The model proposed here is simple and biologically realistic in capturing the major features of volvocine green algae biology 20 , 26–28 that are relevant for the germ-soma differentiation ., The model does not account for the gls genes introducing asymmetry in size between proto-germ and proto-soma cells , but asymmetric division was a late , lineage-specific step in volvocine evolution 44 ., The results presented clearly show that fitness advantages of the division of labor in the presense of strong genetic relatedness of cells in a colony are sufficient to drive the complete differentiation of cells 2 , provided mutations that altruistically remove lineages from the germ line are expressed conditionally 10 , 45 ., Conditionally expressed genes allow the benefits of altruism to go to cells that possess , but do not express , the same allele 10 ., In the model , cell differentiation and the division of labor are driven by individual selection maximizing the number of colony-producing offspring of a colony-producing cell ., That is , the transition to individuality can be explained in terms of immediate selective advantage to individual replicators 2 ., Note that mutant cells that “cheat” by having increased fertility within colonies will tend to lose in competition at the colony level after they develop their own colonies ., Therefore , the conflict between individual and colony level selection is largely removed ., The division of labor is achieved by using the variation in external and/or internal cell environment as a cue to separate the colonys cells by function and then enhance different functions using different subsets of cells ., The colony size has no significant effect on the model dynamics ., In contrast , in Volvox the degree of differentiation between germ- and soma-like cells does correlate with the colony size 26: species with small colonies ( 8–32 cells ) show no cell differentiation , in species with intermediate colonies ( 64–128 cells ) incomplete germ-soma differentiation is observed , and differentiation is complete in species forming large colonies ( 500–5000 cells ) ., However there is a number of biological factors not included in the model explicitly but acting in real cells and colonies which should result in a positive relationship between the colony size and the degree of differentiation ., First , one can reasonably argue that a sufficiently large colony size is necessary for the existence of sufficiently strong gradients in the external environment to which the regulatory genes can react to ., Second , increasing the colony size should result in some spatial heterogeneity between cells in their ability to perform different functions ., For example , inner-layer cells are likely to be less important in contributing towards the colony motility than the outer-layer cells ., Such heterogeneity should decrease the cost of loosing certain functions for some parts of the colony and make the evolution of cell differentiation easier ., Third , the total number of cells performing a particular function in very small colonies may be too small to guarantee an appropriate level of performance especially if the probability of breakage per cell is not small ., A potentially important role for developmental plasticity in the evolution of differentiated multicellularity was emphasized earlier by Schlichting ( 46; see also 29 ) but from a different perspective ., Schlichtings argument was that cell differentiation started as a by-product of random environmental effects translated into new phenotypic forms via pre-existing reaction norms ., Then later favorable phenotypic differentiation became canalized and stabilized via genetic assimilation process ., In contrast , in the scenario considered here developmental plasticity is absent initially and emerges later as a direct result of selection ., Few additional points and connections are worth to be made ., First , the model assumes the existence of undifferentiated multicellular colonies ., Undifferentiated multicellularity has a number of advantages ( e . g . size related ) over single-celled organization and is expected to evolve relatively easily 3 , 16 , 18 , 47 , 48 ., Second , empirical data show a strong positive relationship between the number of cells in an organism and a number of cell types 5 , 49 , 50 ., The classical explanation of this pattern is that increasing the number of cells changes fitness landscape ( e . g . due to physical constraints ) in such a way that differentiation and specialization become necessary for optimizing the efficiency of organisms 5 , 49 , 51 ., In our simple model , the fitness landscape is unaffected by the number of cells in the colony so the model in its current form cannot be used for addressing the question about the relationsips between the number of cells and cell types ., Third , the model is also relevant to ongoing work and discussions on the importance and evolution of modularity , i . e . the separability of the design into units that perform independently , at least to a first approximation 52–54 ., Although there is an emerging agreement that organisms have a modular organization , one of the major open questions is whether modules arise through the action of natural selection or because of biased mutational mechanisms 53 ., In the model considered here , the modules ( e . g . germ and soma ) clearly emerge as a result of selection for reduced fitness trade-offs ., Finally , I should mention some parallels between the models structure and dynamics and the arguments on “groundplans” 55–57 according to which the patterns of labor division in complex organisms and societies are built upon simple changes in the regulation of conserved ancestral genes affecting reproductive physiology and behavior ., The model presented here is expandable in a number of directions including the emergence of multiple cell types , complex organs , or casts of eusocial insects ., For example , the emergence of multiple cell types can be modeled by considering additional cell functions and introducing additional regulatory genes ., The evolution of casts of eusocial insects can be explored by explicitly accounting for regulatory genes that react to the external stumuli ( e . g , food level or pheromones ) affected by the colonys composition ., The majority of existing models of the division of labor in eusocial insects focus on individual worker flexibility in task performance 58 , 59 ., In contrast , the approach introduced here concentrates exclusively on genetically predetermined roles that do not change in time ., Note that genetic variation present in some insect colonies ( e . g . due to polyandry , 60 ) will result in reduced genetic relatedness and , thus , is expected to make conditions for the evolution of the division of labor more strict ., The main result that complete cell differentiation evolves relatively easily and fast supports the view that the transition to differentiated multicellularity , which has happened at least two dozen times in the history of life , is in a sense actually a minor major transition 3 , 8 , 61 , 62 ., It is natural to define fitness as the expected number of offspring colonies in the next generation for a cell starting a colony ., Then , for a cell characterized by viability and fertility , fitness is ( 1 ) In the model , the number of colonies of cells with viability and fertility changes approximately according toTherefore the number of colonies evolves towards the carrying capacity ( 2 ) Assuming that the ecological dynamics ( i . e . changes in the population size ) occur on the faster times scale than the evolutionary dynamics , the ( invasion ) fitness of a mutant cell in a resident population is given by eq . 1 with given by eq ., 2 corresponding to the resident population ., Simplifying , where the approximation is good only if ., Note that the derivative of the invasion fitness function ( with respect to a particular independent variable ) evaluated at the resident population values can be written as With only major gene effects and evolving ( and minor gene effects and set at zero ) , the corresponding invasion fitness gradients areAt an equilibrium ( i . e . , at a singularity ) , ., From the first equation , it follows that at equilibrium and that as ., From the second equation , it follows that at equilibrium and that as ., Eliminating the term from the equalities , one finds that at equilibriumwhich is greater than for ., If , then with given by a solution of equation which simplifies toNote that stays above decreasing to it only asymptotically as ., If , the equilibrium values of and can still be found numerically from the above system of equations ., In the general model , fertility and viability of a monomorphic colony can be written aswhere and are fertilities and and are viabilities of the proto-germ and proto-soma cells ( as defined in Table 1 ) , and is the proportion of proto-germ cells in the colony ., Multidimensional invasion analysis requires one to consider four invasion fitness gradients: and ., Some analytical progress can be achieved if the colony size is very large ( ) ., Under this condition , both major locus effects evolve to ( see the previous subsection ) ., Then we can study the stability of the equilibrium with no gene regulation ( i . e . , with minor locus effect ) to introduction of mutants with small and ., The corresponding invasion fitness gradients are approximated by equations linear in and :where ., Assuming equal genetic variation maintained in both genes , standard linear stability analysis shows that an equilibrium with no gene regulation is locally unstable ifand is stable otherwise ., Figure 4 in the main text illustrates this result ., By considering the four invasion fitness gradients simultaneously ( while still assuming that ) , one can show that if , there exists a singular point at which and ., This suggests that if costs of developmental plasticity are not too big ( i . e . , if , then maximum possible gene suppression evolves ( ) ., Overwise , the minor gene effects stay at intermediate values ( i . e . , between 0 and 1 ) ., Note that with and , the predicted values of and are which is very close to the values observed in numerical simulations with ( see the legend of Figure 4 ) ., Unfortunately , similar simple approach cannot be used for an arbitrary because the equilibrium values of the major locus effects cannot be found explicitly ., In numerical simulations I used all possible combinations of the following parameters: fitness trade-off coefficients , costs of developmental plasticity ; mutation rates ; number of divisions ( so that the colony size was ) ; proportion of the proto-germ cells ., Mutational standard deviation was set to ., The maximum carrying capacity was chosen so that the population with no developmental plasticity ( i . e . with ) evolved to a state at which the number of colonies was close to ., For example , with , was set to , and for and , respectively ., First , I run the model 3 times for each parameter combination each for generations ., Then for parameter values resulting in no differentiation , I did one additional run for generations ., A gallery of numerical results can be viewed in Supporting Information ( Text S1 and Figures S1 , S2 , S3 , S4 , S5 , S6 , S7 , and S8 ) . | Introduction, Results, Discussion, Methods | A crucial step in several major evolutionary transitions is the division of labor between components of the emerging higher-level evolutionary unit ., Examples include the separation of germ and soma in simple multicellular organisms , appearance of multiple cell types and organs in more complex organisms , and emergence of casts in eusocial insects ., How the division of labor was achieved in the face of selfishness of lower-level units is controversial ., I present a simple mathematical model describing the evolutionary emergence of the division of labor via developmental plasticity starting with a colony of undifferentiated cells and ending with completely differentiated multicellular organisms ., I explore how the plausibility and the dynamics of the division of labor depend on its fitness advantage , mutation rate , costs of developmental plasticity , and the colony size ., The model shows that the transition to differentiated multicellularity , which has happened many times in the history of life , can be achieved relatively easily ., My approach is expandable in a number of directions including the emergence of multiple cell types , complex organs , or casts of eusocial insects . | Biological organisms are highly complex and are comprised of many different parts that function to ensure the survival and reproduction of the whole ., How and why the complexity has increased in the course of evolution is a question of great scientific and philosophical significance ., Biologists have identified a number of major transitions in the evolution of complexity including the origin of chromosomes , eukaryotes , sex , multicellular organisms , and social groups in insects ., A crucial step in many of these transitions is the division of labor between components of the emerging higher-level evolutionary unit ., How the division of labor was achieved in the face of selfishness of lower-level units is controversial ., Here I study the emergence of differentiated cell colonies in which one part of the colonys cells ( germ ) specializes in reproduction and the other part of the colonys cells ( soma ) specializes in survival ., Using a mathematical model I show that complete germ-soma differentiation can be achieved relatively easily and fast ( with a million generations ) via the evolution of developmental plasticity ., My approach is expandable in a number of directions including the emergence of multiple cell types , complex organs , or casts of eusocial insects . | evolutionary biology | null |
journal.ppat.1001173 | 2,010 | Cryo Electron Tomography of Native HIV-1 Budding Sites | HIV-1 particles are assembled at the cell membrane , as the 55 kDa viral polyprotein Gag multimerizes on its inner face 1 ., Gag recruits other viral components such as the RNA genome and the surface spike proteins , as well as cellular proteins of the ESCRT machinery required for virus release 2 , 3 , 4 ., The viral protease ( PR ) is essential to convert the immature form of the virion into an infectious mature particle ., Both forms of the virion are pleiomorphic structures , with the repetitive structural elements of the virus arranged non-symmetrically and variably from one particle to the other ., In the immature virion , uncleaved Gag is anchored to the plasma membrane via a charged surface and a myristoyl tail in its N-terminal matrix ( MA ) domain 1 ., As shown by cryo electron microscopy ( cEM ) , Gag arranges in a regular manner , with its internal capsid ( CA ) domain forming a hexameric lattice with a spacing of 8 . 0 nm 5 ., C-terminally of CA , the nucleocapsid ( NC ) domain binds the RNA genome , and the p6 domain recruits the ESCRT machinery to facilitate particle release 6 , 7 ., CA and NC , as well as NC and p6 , are separated by short spacer peptides ( SP1 and SP2 , respectively ) which are processed during maturation ., Proteolytic maturation of HIV-1 has been proposed to initiate at or shortly after assembly and release 8 ., The active dimeric form of the viral PR cleaves Gag and GagPol at multiple sites , leading to the structural transition from the immature particle with its Gag shell forming a truncated sphere to the mature particle with its cone-shaped CA core encasing the condensed nucleoprotein complex in the interior of the virion ., In an in vitro study Pettit et al . detected large differences in processing kinetics at the five cleavage sites in Gag , the cleavage at the site between SP1 and NC being an order of magnitude faster than the second fastest cleavage event 9 ., These results suggested ordered processing during virion maturation , and this was supported by mutagenesis studies of individual cleavage sites 10 ., Processing at all sites except NC-SP1 11 appears to be important for infectivity and small amounts of partially processed Gag products have been shown to exhibit a strong trans-dominant negative effect on viral infectivity 12 , 13 , 14 ., Furthermore , premature processing 15 as well as low concentrations of protease inhibitors , insufficient to significantly affect proteolytic maturation 14 , 16 , 17 , were both shown to efficiently block viral infectivity , indicating an intricate interplay between proteolytic activation and virus formation ., Morphological maturation is believed to occur rapidly following release of the immature virion as no intermediates of maturation have been observed so far and all cellular budding sites appear to carry an immature Gag shell ., The last years have seen increasingly detailed structural studies of released HIV-1 particles in their immature and mature forms 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 ., Analysis of immature virions by cryo electron tomography ( cET ) revealed that the immature lattice covers only part of the viral membrane 22 , arranging as an incomplete “∼2/3” sphere 26 ., These studies suggested that ESCRT involvement in HIV-1 release occurs earlier than previously thought and that so-called late budding sites carrying an almost complete Gag lattice are likely to be dead-end products rather than intermediates in the release pathway 26 ., A later analysis by cET and subtomogram averaging provided a 17 Å structure of the immature Gag lattice and showed how this incomplete hexagonal lattice attains its curvature through inclusion of symmetry defects of irregular shape and size 19 ., In the mature capsid , CA also forms a hexameric lattice , but with a different lattice constant and arrangement than in the immature lattice 27 ., Curvature of the mature capsid is achieved by asymmetric incorporation of pentameric defects at the narrow and wide end of the cone 28 ., While cET studies of released virions have significantly advanced our understanding of HIV-1 morphogenesis , we are currently lacking any three-dimensional information regarding native intracellular particle assembly and budding ., These processes have hitherto been inaccessible for detailed structural studies due to the lack of an experimental system enabling their visualization by cET ., Here , we present a structural study of HIV-1 assembly sites by cET of intact , plunge-frozen human cells ., These snapshots allow detailed structural interpretations of the Gag lattice structure at its assembly site in situ , as well as the arrangement of the cortical actin in its immediate vicinity ., We further determined the structure of a previously not reported Gag lattice type lacking the NC-RNA layer which we propose to be associated with loss of control of PR activation ., To enable structural analysis of HIV-1 assembly in its cellular context , we set out to establish a system for studying this process by cET on intact human cells ., Since cET is limited to samples thinner than ∼500 nm , and thus to thin peripheral areas of adherent cells , the natural HIV-1 host cell types ( T-cells and macrophages ) cannot be used for cET ., This led us to use two human glioblastoma cell lines , U-87 MG and U-373 MG ., These cells grow adherently , have extended , thin peripheral areas , and are amenable to transfection or transduction with adenoviral vectors ., Transfection efficiency proved insufficient for structural analyses , however , and we therefore constructed adenoviral vectors expressing HIV-1 Gag ( AdGag ) and HIV-1 Gag-Pol ( AdGagPol ) , respectively ., Both adenoviral vectors were based on Rev-independent versions of the respective genes 29 , allowing expression of either Gag or Gag-Pol without co-transduction of Rev . U-87 MG or U-373 MG cells were seeded on EM sample grids concomitant with transduction using the respective adenoviral vector ., Immunofluorescence staining for HIV-1 CA at 24–48 h post transduction revealed that almost 100% of the cells expressed HIV-1 Gag , while retaining normal adherent morphology ( Fig . 1A–B ) ., Furthermore , when U-373 MG cells were transduced with AdGagPol , specific cleavage products of HIV-1 Gag were found in the cell lysate and in the particle fraction purified from the culture medium ( Fig . 1C ) , while particle release but no processing was observed for AdGag ( data not shown ) ., Cells transduced with AdGag or AdGagPol were vitrified by plunge-freezing and subjected to cEM ., Upon inspection of the vitrified grids , HIV-1 assembly sites and released virions were present at low but reproducible frequencies at the periphery of the cells ( Fig . 1D ) in regions accessible to cET ., A total of 40 tilt series were recorded on such positions of AdGag and AdGagPol infected cells , respectively ( Table 1 ) ., In cryo electron tomograms of AdGag or AdGagPol transduced cells ( Fig . 2 ) structural features of the cytoplasm were preserved at the high level previously reported for cellular cET 30 , with single ribosomes , actin filaments and microtubules being clearly resolved ., Both viral budding sites ( Fig . 2A , C–F ) and released virions ( Fig . 2B–C , E ) were present in the recorded tomograms ., The budding sites were found at the plasma membrane on lamellipodia–like areas ( Fig . 2C , E ) , occasionally in membrane invaginations ( Fig . 2B , D ) or at the tips ( Fig . 2F , 3B ) or sides ( Fig . 2A , 3A ) of filopodial structures ., In the immature Gag lattice of these budding sites and released particles , the two density layers corresponding to CA and NC-RNA could be resolved , and occasionally the hexagonal symmetry in the CA layer was apparent by eye in more tangential tomographic slices ., Released immature particles ( ip ) were found adjacent to cells expressing Gag ( Fig . 2B , D ) or GagPol ( Fig . 2E ) , whereas released particles with mature morphology ( mp ) were only found adjacent to cells expressing GagPol ( Fig . 2E ) ., No density attributable to components of the ESCRT complex was detected at budding sites at the current resolution ., The preservation of cytoplasmic structure in cET allowed for three-dimensional snapshots of filamentous actin associated with HIV-1 budding sites ., The presence of actin filaments was generally high at budding sites ( Fig . 2A , C , F and Fig . 3A–D ) , as judged by visual inspection where F-actin is clearly resolvable from e . g . intermediate filaments ., Furthermore , we often observed what appeared to be a direct interaction of actin filaments with the budding site ( Fig . 2A , F; Fig . 3D ) ., To categorize the budding sites with respect to their actin context , they were sorted into 5 classes according to the type of actin structures they were associated with ( Fig . 3 , top panel ) ., This classification revealed that 34 of the 39 budding sites analyzed were found adjacent to filamentous actin ( Fig . 3A–D ) , with half of the buds ( 20 of 39 ) appearing on the sides or tips of filopodia-like structures characterized by a parallel actin organization ( Fig . 3A–B ) ., Cryo electron tomograms containing budding sites on intact cells with apparently high signal-to-noise ratio were selected for further analysis by sub-tomogram averaging as previously performed for isolated immature HIV-1 19 ., This analysis provides two kinds of information about the Gag protein lattice: the global arrangement of the repetitive elements into an ordered lattice structure ( “lattice map” ) and the local structure of these repetitive elements ( “unit cell structure” ) ., The immature Gag lattice parameters in the budding sites were the same as previously determined for released immature virions 5 , 19: a hexagonal lattice with a lattice constant of 8 . 0 nm ., The lattice maps derived from the budding sites ( Fig . 4A ) revealed several sites of symmetry breakage , similarly heterogeneous in size and shape as those described for the released virions 19 ., The unit cell structure derived from the cellular tomograms contained density corresponding to the N-terminal and C-terminal domains of CA , respectively , as well as density for the membrane and the NC-RNA layer ( Fig . 5A ) ., At the present resolution ( roughly 40 Å by the 0 . 5 Fourier shell correlation criterion ) , this structure was indistinguishable from that previously reported for released particles 19 ., Immunoblot analysis of AdGagPol-transduced cells and the particle fraction recovered from the culture medium showed proteolytic processing of Gag ( Fig . 1C ) ., Furthermore , particles containing mature-looking cores were found in cryo electron tomograms in immediate proximity of cellular budding sites and next to immature particles ( Fig . 2E ) ., Of note , structures resembling the mature capsid were never found in budding sites still connected to the plasma membrane ., Besides the well-described mature and immature particles , we observed a third , previously not described form of particles in the tomograms of AdGagPol-transduced cells ( Fig . 6A ) ., This apparently “intermediate morphology” exhibiting a thinner Gag lattice than the immature structure was observed both in extracellular particles ( Fig . 6A , Fig . S2 in Supporting Information S1 ) and in HIV-1 budding sites connected to the cell ( Fig . 6A–B ) ., Budding sites and extracellular particles with the previously unrecognized Gag lattice were characterized by a single density layer bound to the inner face of the membrane ( Figure 6B ) , as opposed to the two layers ( CA and NC-RNA ) seen in the immature lattice ( Fig . 6C ) ., Density plots orthogonal to the cell membrane confirmed that the density layer found in these novel structures was located at the same position relative to the plasma membrane ( peak at −12 nm ) as the CA layer of the complete immature lattice ( Fig . S1 in Supporting Information S1 ) ., Strikingly , budding sites and released particles with this novel Gag lattice displayed the same hexagonal symmetry and lattice spacing as the immature lattice ( Fig . 4B ) ., The apparently intermediate type lattices were typically more complete with smaller lattice defects than the immature lattices , but the lattice defects were similarly heterogeneous ., Furthermore , the unit cell structure of the new lattice revealed an arrangement identical to the CA region of the immature lattice , with density corresponding to the N- and C-terminal domains of CA ( Fig . 5B ) , but lacking any density corresponding to the NC-RNA layer ., The cellular budding sites and extracellular particles containing this lattice also lacked the internal density corresponding to a condensed NC-RNA complex ., Such densities were readily observed within mature capsids and in virions carrying partially processed Gag proteins ( de Marco et al . , accompanying paper ) ., To determine whether the newly described lattice also occurs in other HIV-1 producing cells including T-cell lines , we recorded electron tomograms on sections of resin-embedded cells producing HIV-1 ., Although resin-embedded material is not preserved at the molecular level , it might still be possible to distinguish the two lattice types in these data , and this would allow recording sufficiently large data sets for statistical analyses ., Visual inspection of resin-embedded sections from infected MT4 cells suggested the presence of a thinner density layer in some budding sites ( Fig . 7A , B ) ., This interpretation was validated based on the following criteria:, ( i ) different types of membrane-bound densities in budding sites in resin embeddings could be classified by an unbiased scheme , and, ( ii ) the presence of structures corresponding to the novel Gag lattice was dependent on an active viral PR ., Classification of budding sites and particles was based on linear density plots of the membrane-bound Gag density ., Fig . 7C shows a typical example , demonstrating that enough structure is retained to resolve the plasma membrane and distinct regions in the Gag layer ., The classification was based on the parts of the plots containing the density bound to the inner face of the membrane ( shaded area in Fig . 7C ) ., Data were recorded on resin embeddings of HIV-1 infected MT4 cells , HeLa cells transfected with a proviral HIV-1 plasmid as well as positive and negative controls for the presence of the two lattice types ( Table 2 ) ., In order to avoid overrepresentation of single cells , two tilt series were recorded per cell , and they were recorded as far apart as possible in the section plane ( usually >5 µm ) ., The same cell was never studied in consecutive sections ., Our published data on infected MT4 cells 26 were also included in the analysis ( Table 2 , data set 8 ) ., Principal component analysis of the variance in the data set revealed the presence of one major cluster with residual variance and a smaller , more homogeneous one ( Fig . S3A–D in Supporting Information S1 ) ., Hierarchical clustering of the cropped line plots identified the two main classes as the two clusters observed in the PCA representation ( Fig . S3E , F in Supporting Information S1 ) ., The less populated class ( class 1 , dashed line in Fig . 7D ) lacked the innermost density peak corresponding to NC-RNA , whereas this peak was present in the more populated class ( class 2 , solid line in Fig . 7D ) ., Further division of class 2 into subclasses mainly revealed a slight variation in the position and relative intensity of the NC-RNA peak ( Fig . S4 in Supporting Information S1 ) ., Table 3 shows the distribution of budding sites into the two classes for each sample ., Class 1 contained 18% of the budding sites on HIV-1 infected MT4 cells and 2 . 9% of the budding sites on transfected HeLa cells ., For infected MT4 cells treated with the PR inhibitor lopinavir , no class 1 budding sites were observed ., As a control for the thinner Gag lattice , released virions carrying uncleaved MA-SP1 were studied ., In this case , Gag processing is stalled at the MACASP1 fragment due to inactivating mutations at cleavage sites between MA and CA and CA and SP1 , respectively ( see 10 and accompanying paper by de Marco et al . ) ., These particles generally fell into class 1 ., Taken together , these results suggest that class 1 corresponds to the newly described lattice detected by cET , while class 2 corresponds to the immature lattice ., Radius and closure of the Gag layer was calculated separately for the two classes ( Table 4 , Fig . 7F ) ., Whereas the average radii of the two classes were similar , the closure of the Gag layer was significantly higher for the class 1 buds ( Wilcoxon rank sum test , p\u200a=\u200a0 . 0021 ) ., Assuming the same lattice packing in both classes , class 1 budding sites in this data set contain 5 , 000±1 , 300 Gag molecules while class 2 budding sites in this data set contain 3 , 300±1 , 500 Gag molecules ., Finally , we asked whether individual cells carried only one class of budding sites or both ., In the case of HIV-1 infected MT4 cells , 47 cells exhibited more than one budding site in the tomogram ( Table 2 , data set 4 ) ., Fig . 7E ( black bars ) shows the relative frequency of class 1 budding sites on these 47 cells , overlayed with the theoretical distribution ( grey circles ) that would be expected if all cells had the average of 18% class 1 budding sites ., Clearly , most cells had either only class 1 or only class 2 budding sites , indicating that the budding phenotype is determined at the single cell level ., Here , we established a system to visualize HIV-1 virus-like particle assembly and release in situ at macromolecular resolution in three dimensions using cET ., This method permits structural interpretation of assembly sites and their cytoplasmic and membrane surroundings without the limitations imposed by the fixation , staining and dehydration methods commonly used in cellular electron microscopy ., It allowed a detailed analysis of the Gag lattice in cellular budding sites , suggesting that the organization of the immature HIV-1 particle is already determined at the point of its intracellular assembly , and further identified a hitherto undescribed Gag lattice form ., We show that this Gag lattice form is predominant in a subset of infected T-cells , and hypothesize that it is the result of premature PR activation in these cells ., The cET analysis of HIV-1 Gag and Gag-Pol expressing cells also provided unprecedented snapshots of cortical actin filaments in the vicinity of viral budding sites ( Fig . 2 , 3 ) ., It is well established that HIV-1 particles contain substantial amounts of actin 31 , but the interplay between cortical actin and retrovirus assembly is still unknown ., A recent study indicated a role for star-shaped actin filament arrangements emanating from the viral budding sites in HIV-1 assembly 32 ., Our cET data provide 3D structures suggesting directed arrangement of actin filaments towards some budding sites , particularly in filopodia-associated buds ., Strikingly , half of the budding sites in the current data set were present on actin-filled filopodia ., To what extent this is a consequence of the particular region of the cell accessible to cET remains to be determined ., Future cET studies using altered Gag constructs and selective manipulation of the actin dynamics will provide the necessary tools to untangle the molecular details of the involvement of actin at HIV-1 budding sites ., Extracellular immature HIV-1 particles are the direct products of viral budding at the plasma membrane , making it likely that their Gag protein lattice is determined during assembly ., The observation that the Gag lattice is incomplete in the immature virus 22 , 26 and contains defects 19 , 22 raised the alternative hypothesis that Gag rearrangements occur following release ., This could involve post-release association of smaller patches of Gag hexamers , or post-release symmetry breakage and dissociation of a hexagonal lattice containing evenly spaced pentameric defects 22 ., Analysis of the lattice in extracellular immature virions revealed continuous hexameric symmetry over most of the lattice with curvature induced by defects of irregular size and shape 19 ., These studies were performed on purified released particles , however , and the degree of order attained during assembly at the plasma membrane may have been higher ( and broken after release ) or lower ( and subsequently assembled into a more ordered lattice ) ., Our cET data of budding sites revealed structures of the immature Gag lattice in intact unperturbed cells prior to release , allowing us to draw conclusions on the early steps of the HIV-1 assembly process ., There was no indication of separated “islands” of hexameric Gag in the in situ budding sites; upon visual inspection they consistently contained one continuous layer of Gag protein ., A more detailed analysis of the lattice in the immature budding sites revealed that its unit cell structure was highly similar to that previously described for released immature virions 19 , and included irregular lattice defects of similar size and distribution ., This suggests that the organization of the immature HIV-1 particle is indeed determined at the point of its intracellular assembly , and not as a result of large scale post-release ordering or disordering ., Besides the described immature Gag lattice , we observed a previously undescribed lattice type in extracellular particles and in budding sites still connected to the cell ., Formation of this lattice was dependent on an active HIV-1 PR ., The newly discovered lattice exhibited hexagonal symmetry with a CA layer identical to the immature lattice , but lacked the second layer of density corresponding to the NC-RNA complex ., The particles and budding sites also contained no visible NC-RNA condensate in the center of the particles or adjacent to the buds , suggesting that the viral genome is absent ., A comparison with the analysis performed in the accompanying paper by de Marco et al . on isolated virions with specific Gag cleavage site mutations showed that the newly described lattice is most likely composed of the MA-SP1 region of HIV-1 Gag ., Although the tomographic average at current resolution does not allow identification of the density corresponding to the SP1 peptide , de Marco et al . showed that this peptide is present , independent of whether the cleavage site between CA and SP1 is intact or not ., Notably , the particles produced from the MA-CA and MA-SP1 cleavage site mutants which exhibit an MA-SP1 type lattice do contain a clearly distinguishable density corresponding to condensed NC-RNA ( de Marco et al . ) , which distinguishes them from the particles with this lattice type in the current study ., As further described in the accompanying paper by de Marco et al . , the MACASP1 lattice is likely to constitute a natural intermediate in the proteolytic maturation of HIV-1 ., Cleavage between SP1 and NC is the fastest processing step in vitro and abscission of the C-terminal NC-p6 regions from the immature lattice is required for condensation of the inner ribonucleoprotein complex of the virion prior to formation of the capsid ., Thus , extracellular particles containing the MACASP1 lattice and inner density corresponding to NC-RNA are suggested to represent a “frozen” intermediate in HIV-1 maturation ( de Marco et al . ) ., Such particles were not observed for wild-type HIV-1 constructs , however , indicating that maturation intermediates are normally short-lived ., An MACASP1 lattice without adjacent NC-RNA density was observed in 18% of HIV-1 budding sites on infected MT-4 cells , on the other hand , and its frequent presence suggests that it is a metastable structure with a significant lifetime in this case ., We previously reported that budding sites on infected MT4 cells have a broad distribution of Gag layer morphologies ., Some resemble actual released immature virions , but some have a more complete Gag layer , containing more Gag molecules than the released virions and thus rather resemble arrested “late” budding sites not thought to be virus precursors 26 ., When re-evaluating these data with respect to the type of Gag lattice , the immature lattice was predominantly found in budding sites resembling released virions , whereas the budding sites with the novel lattice had a significantly more closed Gag layer than released virions ( Table 4 , Fig . 7F ) ., In the case of late-domain defective variants , formation of a closed layer is thought to be caused by a failure to recruit the ESCRT machinery and thus to drive the release process ., Consequently , Gag assembly continues until reaching an equilibrium structure , but the resulting late budding sites are mostly dead-end products ., It appears likely that a similar process is also relevant for the budding sites with the novel lattice ., We hypothesize that premature proteolytic activation prior to confinement of virion constituents in the budding virion leads to removal of NC-p6 from Gag and of PR and downstream pol products from Gag-Pol ., The cleaved C-terminal fragments are lost from the assembly site , while the MACASP1 lattice is stably retained ., No ESCRT recruitment occurs since the p6 domain has been removed and no further processing occurs because PR , not being confined in a viral particle , has been lost ., Accordingly , no budding structures with mature cores were observed ., Further assembly of the truncated lattice may involve additional Gag molecules undergoing partial processing or addition of cytoplasmic MACASP1 molecules since a truncated MACASP1 construct has been shown to form budlike structures when overexpressed 33 , 34 ., The presence of extracellular particles with the MACASP1 lattice and lacking NC-RNA may indicate some ESCRT-independent release ., Release of virions lacking most of the NC domain ( but in this case carrying the p6 domain ) has also been observed in previous studies 35 , 36 ., The results of the current study clearly show that morphological maturation can be initiated inside the producer cell , but indicate that it is largely non-productive if the virion constituents are not confined in a structure sequestered from the cytoplasm ., Assembly , release and maturation would therefore require the sequential processes of Gag association , ESCRT recruitment ( prior to closure of the Gag shell ) , and PR activation ( after or concomitant with ESCRT-mediated sequestration of the bud ) ., The observation of aberrant , apparently dead-end structures in a subset of cells in HIV-1 infected T-cell lines suggests that the control of these processes can be lost , however ., Infected MT-4 cells generally contained only one type of budding sites indicating that this loss of control may be caused by the host cell environment ., Conceivably , host cell factors may be involved in regulating the kinetics of the consecutive stages of viral assembly , budding and release , and the subset of T-cells may have lost this regulation , although a general cytopathic effect cannot be ruled out at present ., Identifying viral or host factors involved in this regulation will be of great importance for our understanding of HIV-1 morphogenesis , and may also lead to new approaches to render HIV-1 infection non-productive ., Adenoviral vectors for HIV-1 Gag ( AdGag ) and GagPol ( AdGagPol ) were constructed from the Rev independent versions of these genes 29 , using the BD Adeno-X Expression System 1 ( BD Biosciences Clontech ) ., Briefly , the 1539 nt gag-gene was amplified by PCR adding flanking NheI/XbaI sites and ligated into the transfer vector pShuttle2 ., The gag-pol gene was excised with NdeI/XbaI from pcDNA3 . 1syngagpol 29 and directly cloned into pShuttle2 ., After verification of the insert by sequence analysis , these plasmids were digested with PI-Sce/I-Ceu ., The inserts were ligated into the pre-digested Adeno . X viral DNA ., Resulting plasmids were verified by restriction digest and DNA sequencing ., Recombinant adenoviruses were generated by transfection of the AdGag or Ad GagPol plasmids into HEK 293 cells and amplified according to the protocol supplied with the kit ., Titers of adenoviral vectors were determined using the Adeno-X rapid Titer Kit ( Clontech ) ; titers of both AdGag and Ad GagPol typically reached 1–2×109/ml ., Immunofluorescence staining of AdGag and AdGagPol transduced cells for HIV-1 Gag was performed using a rabbit polyclonal anti-CA antibody and a FITC-conjugated goat-anti-rabbit secondary antibody ., Western blots of cell lysates and particles purified from the culture media of AdGag and AdGagPol transduced cells were performed using the same polyclonal anti-CA antibody , with an IRDye 800CW goat anti-rabbit secondary antibody ( LI-COR Biosciences ) as recommended by the manufacturer ., Human glioblastoma cell lines U-87 MG and U-373 MG were maintained in DMEM and MEM , respectively , supplemented with Penicillin-Streptomycin and 10% foetal bovine serum ., When grown on EM grids ( CF-2/1-2AU , Protochips , inc . ) , cells were seeded at low density and kept in medium with 0 . 5% serum which increased the extent of flat areas of the cells ., Transduction with AdGag or AdGagPol was performed concomitant with or one day after seeding cells on EM grids ., A multiplicity of ‘infection’ ( MOI ) of 10 was typically used , calculated based on cell culture dish area and assuming that added vector will adhere to any substrate surface ., One to three days after seeding cells on the grids , they were plunge-frozen into liquid ethane after application of 10 nm colloidal gold ., Cell culture and preparation for sections of resin embedded cells were performed as previously described 26 ., For production of immature virus , cells were grown in the presence of 1 µM lopinavir ., For production of particles containing an uncleaved MA-SP1 layer , HEK293T cells were transfected with a proviral HIV-1NL4-3 plasmid carrying mutations in the cleavage sites between MA and CA and CA and SP1 , respectively ( 14 and accompanying paper by de Marco et al . ) ., Cells were fixed , stained , dehydrated and EPON-embedded , and 300 nm sections of the embeddings were cut and transferred to EM grids after deposition of colloidal gold onto the EM grid support film ., Cryo electron tomography was performed with a Philips CM300 or a FEI Tecnai F30 Polara transmission electron microscope ( FEI; Eindhoven , The Netherlands ) , both equipped with 300 kV field emission guns , Gatan GIF 2002 post-column energy filters and 2048×2048 Multiscan CCD cameras ( Gatan; Pleasanton , CA ) ., All data collection was performed at 300 kV , with the energy filter operated in zero loss mode ., Tilt series of cryo specimens were typically recorded from −60° to +60° with an angular increment of 1 . 5° , a total electron dose of 80 electrons Å−2 , and a defocus of −6 . 0 µm to −8 . 0 µm ., The pixel sizes at the specimen level were 8 . 21 Å at the CM300 , and 8 . 05 Å or 7 . 13 Å at the FEI Tecnai F30 Polara , respectively ., Electron tomography of resin sections was performed at an FEI Tecnai F20 transmission electron microscope ( FEI; Eindhoven , The Netherlands ) , equipped with 200 kV field emission gun and a 4096×4096 Eagle CCD camera ( FEI; Eindhoven , The Netherlands ) ., Tilt series on resin sections were recorded at room temperature from −60° to +60° with an angular increment of 1 . 0° , with a one time binned pixel size of 2*3 . 72 Å and a defocus of -4 . 0 µm ., For both cryo and room temperature tilt series , the applied electron dose for a given tilt angle α was proportional to 1/cos ( α ) to compensate for the higher effective specimen thickness at higher tilts ., At the Tecnai microscopes , the SerialEM acquisition software 37 was used for tilt series acquisition ., Three-dimensional reconstructions from tilt series were calculated using the weighted back-projection method 38 , as implemented in the TOM toolbox 39 for MATLAB ( Mathworks , Natick , Massachusetts , United States ) with | Introduction, Results, Discussion, Materials and Methods | The structure of immature and mature HIV-1 particles has been analyzed in detail by cryo electron microscopy , while no such studies have been reported for cellular HIV-1 budding sites ., Here , we established a system for studying HIV-1 virus-like particle assembly and release by cryo electron tomography of intact human cells ., The lattice of the structural Gag protein in budding sites was indistinguishable from that of the released immature virion , suggesting that its organization is determined at the assembly site without major subsequent rearrangements ., Besides the immature lattice , a previously not described Gag lattice was detected in some budding sites and released particles; this lattice was found at high frequencies in a subset of infected T-cells ., It displays the same hexagonal symmetry and spacing in the MA-CA layer as the immature lattice , but lacks density corresponding to NC-RNA-p6 ., Buds and released particles carrying this lattice consistently lacked the viral ribonucleoprotein complex , suggesting that they correspond to aberrant products due to premature proteolytic activation ., We hypothesize that cellular and/or viral factors normally control the onset of proteolytic maturation during assembly and release , and that this control has been lost in a subset of infected T-cells leading to formation of aberrant particles . | The production of new HIV-1 particles is initiated at the plasma membrane where the viral polyprotein Gag assembles into a budding site , and proceeds through release of an immature virion which is subsequently transformed to the infectious virion by proteolytic cleavage of Gag ., Here , we established experimental systems to study HIV-1 budding sites by cryo electron tomography ., This technique allows three-dimensional structure determination of single objects at macromolecular resolution , thus being uniquely suited to study variable structures such as HIV-1 particles and budding sites ., Using cryo electron tomography , we obtained three-dimensional images with unprecedented detail of the formation of HIV-1 particles ., By analyzing these images we show that the organization of released immature HIV-1 is determined at its intracellular assembly without major subsequent rearrangements ., We further identify a lattice structure of the viral protein Gag present in budding sites that seem to lack the viral genome and thus cannot be precursors of infectious viruses ., We show that some HIV-1 infected T-cells preferentially carry these budding sites , suggesting that they have lost a crucial control of the proteolytic maturation of the virus . | virology/virion structure, assembly, and egress, virology/immunodeficiency viruses, biophysics/macromolecular assemblies and machines | null |
journal.pgen.1004585 | 2,014 | Hydroxymethylated Cytosines Are Associated with Elevated C to G Transversion Rates | In mammalian genomes , most cytosines that occur in a CpG context are methylated ., 5-methylcytosines ( 5mCs ) at CpG dinucleotides exhibit mutation rates an order of magnitude above that of unmodified cytosines , a consequence both of their greater propensity to deaminate and error-prone repair of the resulting thymine 1 ., This mutational liability is evident in higher levels of single nucleotide polymorphisms ( SNPs ) segregating at CpGs in mammalian populations 2–4 , higher rates of divergence between species at these sites 5 , 6 , and higher somatic mutation rates in many cancer genomes compared to other nucleotide contexts 7 ., Recently , it has become clear that the repertoire of naturally occurring cytosine modifications in mammals extends beyond 5mC to include a series of modifications derived from successive rounds of 5mC oxidation: 5-hydroxymethylcytosine ( 5hmC ) , 5-formylcytosine ( 5fC ) , and 5-carboxylcytosine ( 5caC ) 8 , 9 ., 5fC and 5caC have been found to occur at low frequencies in genome-wide studies in human and mouse ( ∼0 . 01–0 . 0001% of cytosines 10 ) , consistent with being rapidly converted intermediates in an active demethylation pathway that involves cumulative oxidation of 5mC by Tet enzymes and the eventual removal of 5fC or 5caC via base excision repair ( BER ) 11 ., In contrast , 5hmC has been detected at relatively high levels ( ∼0 . 1% of cytosines ) in certain cell types including Purkinje cells , embryonic stem ( ES ) cells and primordial germ cells , suggesting that it might be present as a quasi-stable epigenetic mark rather than merely a transient demethylation intermediate 12 ., In the context of the high mutational burden of 5mC and considering that 5hmC can be present as a stable epigenetic mark , we wondered whether methylated and hydroxymethylated sites might be associated with distinct patterns of sequence evolution , perhaps as a consequence of divergent mutational biases ., For example , in mammalian systems , repair of 5hmU:G mismatches ( derived from 5hmC deamination ) by the glycosylases TDG and SMUG1 is less error-prone than dealing with 5mC-derived T:G mispairs 13 ., As a consequence , residues that spend a significant proportion of their lifetime in the germline in a 5hmC state might be less mutagenic than 5mC sites ., Here , to elucidate the evolutionary repercussions of hydroxymethylation , we integrate population genomic , inter-species divergence and somatic mutation data from tumors with publicly available base-resolution maps of 5hmC and 5mC in human and mouse ES cells 14–17 ., As further discussed below , 5hmC profiles in ES cells show similarities to 5hmC profiles at different stages of germline development , making ES cells a relevant model system to investigate the impact of hydroxymethylation on sequence evolution ., We then focussed on residues located outside of repeat regions , covered by at least ten sequencing reads in the pertinent bisulfite experiment , and amenable to accurate SNP calling ( Materials and Methods ) ., Further , as cell line genotypes differ from the reference genomes , we confined analysis to sites with known cell line genotype , using ENCODE short read data to genotype H1 hESC ( Materials and Methods ) ., For this high-confidence dataset , we asked what fraction of 5mC , 5hmC , and C sites are associated with a derived SNP ( “SNP rate” ) in the human population and across 17 different laboratory or wild-derived inbred mouse strains ., As shown in Figure 1 , there is a small ( but significant ) reduction in C to T SNP rates at 5hmC compared to 5mC sites , consistent with less error-prone repair of 5hmU compared to T as suggested above ., Unexpectedly , however , 5hmC sites in both human and mouse exhibit substantially higher rates of C to G transversions than 5mC sites , with C to A rates additionally elevated in human ., Regarding the relative frequency of different base changes , transitions are an order of magnitude more common than transversions for both 5mC and 5hmC , likely reflecting high mutation rates following deamination ., Next , we considered rate differences at the level of divergence between species ., For sites inferred to be cytosines in the human-chimp ancestral genome ( see Materials and Methods ) , we examined substitutions along the chimp lineage as a function of methylation state in human ., Consistent with the population genomic data , transversion rates are higher at 5hmC sites ( Figure 1 ) ., Analysis of substitutions in the M . spretus genome – relative to the M . musculus-M ., spretus ancestral genome and M . musculus methylation state – echoes this result: C to G rates are higher at 5hmC sites than at 5mC sites ( Figure 1 ) ., The incidence of 5hmC sites varies according to regional GC content 15 , 19 , functional context ( intron , exon , promoter , etc . ) 20 , and chromatin environment , where it is associated with active transcription and certain enhancer states 21 , 22 ., 5mC and 5hmC sites might therefore exhibit distinct patterns of sequence change not because of intrinsic ( mutational ) differences between the two marks but because they are unevenly represented in functional elements or genomic regions that are governed by disparate mutational and/or selective regimes 23 , 24 ., Indeed , examining derived allele frequencies ( DAFs ) in the human population we find a significant excess of rare alleles at 5mC compared to 5hmC sites ( P<10−20 ) , suggesting stronger average purifying selection at 5mC sites ( Figure S2 ) ., In order to isolate 5hmC/5mC-specific patterns of evolution that are independent of functional context and therefore likely mutational in nature , we adopted the following strategy: for every 5hmC site we selected a 5mC site that matches the 5hmC site with regard to local ( ±50 nt around the focal site ) and regional ( ±500 nt ) GC content , chromatin state , biotype , the upstream neighbouring nucleotide and the methylation level of the focal cytosine ( see Materials and Methods and Table S1 for details ) ., Matching for methylation level is particularly important given previous findings that more highly methylated CpGs in human sperm are associated with a greater frequency of rare derived alleles 4 , consistent with selection being stronger , on average , at highly methylated sites ., Concurrently matching across multiple criteria in this fashion is feasible because 5mC sites vastly outnumber 5hmC sites so that a match can be found for a large fraction of 5hmCs ., We did not include unmethylated cytosines in this analysis because matching across three categories severely reduces sample size ., As mammalian hydroxymethylation occurs almost exclusively at CpG dinucleotides 15 , 20 , we focus on sites in the CpG context ., All rate estimates below , including in the context of tumor evolution , refer to this context ., This matching procedure yields 121604 and 154060 5hmC-5mC pairs for human and mouse , respectively , which are matched with regard to various potential confounders and no longer differ significantly in their DAF spectra ( P\u200a=\u200a0 . 1 , Figure S2 ) , suggesting a similar distribution of selective constraints for the two classes of sites ., Comparing SNP rates across matched sites suggests that differences in C to T rates between 5mC and 5hmC sites are indeed minor , and only remain marginally supported in mouse ( Figure 2A , fold difference in rate ( 5hm/5mc ) : human: 0 . 99; mouse: 0 . 96 ) ., Importantly , however , pronounced differences in C to G transversion rates remain evident in both mouse and human ( fold difference: human: 1 . 43; mouse: 1 . 22 ) ., Moreover , faster C to G rates at 5hmC sites are found across different chromatin states , biotypes ( Figure 2C , Human: P<8*10−6 , Mouse: P<0 . 0005; binomial test , testing for likelihood of all chromatin states showing enrichment in the same direction ) , and GC content levels ( Figure 2D ) and appear independent of the immediate nucleotide context ( Figure 2B ) ., For many of these subsets , differences are individually significant and we do not find a single context where the C to G rate is faster at 5mC sites ., Furthermore , the effect is insensitive to nucleosome occupancy ( Figure 2F ) and observed in both open and closed chromatin as defined by the ENCODE project for H1 hESC ( Figure 2E ) , suggesting that it is not simply a corollary of differential DNA accessibility , with , for example , more open chromatin structure facilitating Tet-mediated 5hmC generation 25 but also rendering DNA more prone to oxidative damage , a cause of C to G transversions 26 ., Having systematically accounted for differences in functional and sequence context , we reasoned that differences between 5mC and 5hmC sites likely reflect mutational biases ., However , any mutational bias model rests on the assumption that ( hydroxy ) methylation patterns in ES cells are predictive of patterns in the germline and can therefore contribute mechanistically to a 5hmC-related mutation signature ., To evaluate this assumption we first considered base resolution 5hmC maps for mouse neurons ( adult frontal cortex ) 17 ., In particular , we focused on sites with evidence for hydroxymethylation in neurons but not in ES cells ., Hydroxymethylation that is present exclusively in differentiated cells such as frontal cortex neurons should have no bearing on mutation dynamics in the germline ., Neuron-specific 5hmC sites should , in mutational terms , behave like germline 5mC sites ., We repeated the matching procedure described above , but now pairing neuron-specific 5hmC sites to sites called as 5mC in both ES cells and neurons ., As predicted , there is no difference in C to G rates between the matched pairs ( Figure 2G ) and rates at neuron-specific 5hmC sites are significantly lower than at 5hmC sites in mESCs ( P\u200a=\u200a0 . 0009 ) ., Importantly , hydroxymethylation is more common in neurons , so this result is not an artefact of reduced power ( number of matched pairs N\u200a=\u200a428032 ) ., The genomic incidence of hydroxymethylation has previously been examined for different stages of mouse spermatogenesis , using a chemical labelling method followed by enrichment and sequencing 27 ., We find that 5hmC sites in ES cells are overrepresented in 5hmC-enriched regions in sperm , particularly at earlier stages of spermatogenesis ( Figure S3 ) ., In addition , at multiple stages of spermatogenesis we find significant differences in C to G SNP rates ( calculated for 5hmC and 5mC sites in ES cells ) in 5hmC-enriched regions ( Figure S3 ) ., In contrast , we never observe significant differences in regions without 5hmC enrichment ., Note that there is high overlap in 5hmC-enriched regions across different stages of spermatogenesis 27 , precluding statistically meaningful analysis of sites exclusively hydroxymethylated at some stages but not others ., Future base resolution data will be required to establish more precisely to what degree hydroxymethylation patterns in the germline and ES cells overlap ., However , based on the data presented and unpublished data showing high levels of similarity between 5hmC profiles in ES cells and the early germline ( P . Hajkova , unpublished results ) , we suggest that ES cells constitute a relevant proxy to study the evolutionary repercussions of hydroxymethylation ., We reasoned that – if elevated C to G rates are mechanistically linked to hydroxymethylation – they might be higher at sites where the 5hmC mark is more prevalent ., Hydroxymethylation is non-stoichiometric and sites classified as 5hmC are typically hydroxymethylated in a minority of cells in the population ., We therefore tested whether cytosines with higher levels of hydroxymethylation exhibit higher SNP rates ., This is indeed the case in human ( Figure 3 , P\u200a=\u200a0 . 04; test of proportions comparing terminal bins ) ., Although an increase towards higher rates for highly hydroxymethylated sites is also apparent in mouse , the difference is not significant ., If differences at 5hmC sites reflect mutational biases , such biases might also operate in the context of somatic evolution ., To explore this possibility , we compiled a catalogue of single nucleotide mutations across 346 diverse fully sequenced cancer genomes ( see Materials and Methods ) and compared somatic mutation rates for the set of matched 5hmC and 5mC sites described above ., Again , we find significantly elevated C to G rates at 5hmC sites ( Figure 1 ) ., We then examined the relationship between C to G rates in tumors and the expression of Tet proteins ., Tet proteins catalyse the oxidation of 5mC to 5hmC and therefore constitute a critical rate-limiting step for 5hmC generation , as evident in lower genome-wide levels of 5hmC in mouse ES cells where Tet1/2 protein levels are diminished following shRNA-mediated knock-down 28 ., As Tet expression levels affect the relative abundance of 5hmC , we predict that Tet expression should positively correlate with C to G mutation rates , irrespective of low baseline hydroxymethylation levels in cancer cells compared to ES cells or neurons ., Figure 4A highlights that , considering mutations across 346 cancer genomes , there are positive correlations between the proportion of all mutations that are C to G ( %C2G ) and the transcript levels of Tet1 and Tet3 ., To ascertain whether correlations are stronger than expected by chance , we compared each Tet gene to a bespoke control set of ∼1500 genes most similar in median expression and dispersion across tumors ( see Materials and Methods ) ., As some mutational processes that operate in cancer genomes are known to exhibit nucleotide context biases 7 , we present correlation coefficients separately for each upstream neighbouring nucleotide ., The results confirm that expression levels for Tet1 and Tet3 , but not Tet2 , are strongly associated with %C2G ( Figure 4B , Tet1: P<1 . 57*10−05; Tet2: P>0 . 05; Tet3: P<4 . 58*10−07; Stouffer test combining P values across contexts ) and largely insensitive to upstream nucleotide context , suggesting that we are not dealing with a known , context-dependent mutational process ., Considering correlations separately for 11 different types of cancer ( colorectal cancers , breast cancers , etc . ; see Table S2 for a complete list ) , we also predominantly observe positive correlations for Tet1 ( 35 out of 44 cancer type-context combinations ) and Tet3 ( 32/44 ) but not for Tet2 ( 17/44 , Figure 4C ) ., In terms of the variance explained by Tet expression levels , correlations are comparable in magnitude to the correlation between APOBEC signature mutations and APOBEC expression recently reported for breast cancer genomes 29 ., To probe further into the putative link between Tet activity , hydroxymethylation , and C to G transversions , we considered SNP rates in relation to Tet1 binding footprints , determined on a genome-wide scale in mouse ES cells 30 ., Although coinciding surprisingly poorly with the distribution of 5hmC sites 30 , 31 , we reasoned that Tet1 binding can be exploited as a sentinel for intrinsic hydroxymethylation risk alongside 5hmC/5mC status itself ., 5mC sites can be seen as refractory to hydroxymethylation if they are located inside a Tet1 binding footprint yet fail to show signs of hydroxymethylation ., Conversely , 5hmC residues located in Tet1 binding footprints clearly can be hydroxymethylated and are likely hydroxymethylated more reproducibly across cells and time given the presence of Tet1 ., On average , 5hmC sites inside Tet1 binding footprints should therefore spend more time in a hydroxymethylated state than 5hmC sites outside footprints ., In line with this scenario , we observe the highest and lowest rate of C to G transversions at 5hmC and 5mC sites inside Tet1 binding footprints , respectively ( Figure 4D ) ., This finding also argues against a scenario where elevated transversion rates are simply the consequence of a locally elevated non-specific oxidation risk associated with the presence of Tet proteins ., If different mutational dynamics at 5hmC sites are associated with Tet-mediated oxidation , we might also suspect regions of high 5hmC turnover – where 5hmC is frequently further oxidized to 5fC/5caC and eventually undergoes BER – to show more pronounced rate differences ., Considering the presence of 5fC as an indicator of high 5hmC turnover , we compared SNP rates inside and outside regions found to be enriched for 5fC in mESC 32 ., We observe trends in the expected direction for all base changes , with C to G rates more pronounced for sites located in 5fC-enriched regions ( Figure 4E ) ., However , because there are few 5fC-enriched regions and therefore few nucleotides available for analysis , SNP rate estimates are correspondingly noisy , likely precluding the detection of a significant differences between 5hmC sites and residues located in 5fC-enriched regions ., Yu and colleagues characterized hydroxymethylation as predominantly asymmetric - that is , at CpG dinucleotides where one cytosine showed evidence for hydroxymethylation , the cytosine on the opposite strand typically did not 15 ., In contrast , 5mC sites are highly symmetric , with 99% of CpG dinucleotides – when methylated – methylated on both strands 16 ., Although 5hmC asymmetry might to some extent be owing to low sequencing depth 20 , several high resolution studies now support asymmetric hydroxymethylation as a genuine phenomenon 15 , 33 , 34 ., Indeed , asymmetric hydroxymethylation must occur temporarily given that Tet enzymes oxidize a single 5mC site at a time 35 ., We therefore examined SNP rates at symmetrically and asymmetrically hydroxymethylated CpGs ., Because this analysis requires consideration of consecutive cytosines on opposite strands , we use the total pool of eligible CpG dinucleotides rather than the matched set employed previously ., In both human and mouse , rates of cytosine loss at 5hmC sites appear consistently higher when the 5hmC is found in an asymmetric context ( Figure 5A , P<0 . 04 , binomial test , testing for consistency of enrichment across mutations and species ) ., Note that symmetrically hydroxymethylated sites are rare , so our power to detect differences for transversions is limited ., We demonstrate here that hydroxymethylated cytosines in human or mouse ES cells show different patterns of sequence variation and evolution compared to their 5mC-methylated counterparts ., They are more likely to give rise to C to G transversions segregating in the population , more frequently associated with C to G substitutions in closely related sister species and exhibit higher rates of C to G mutations in tumors ., As rates correlate with quantitative levels of 5hmC , Tet expression/binding , and the presence of 5fC , we suggest that rate differences between 5hmC and 5mC sites – consistently observed across different functional and sequences contexts – are likely mutational in origin and mechanistically linked to hydroxymethylation rather than the result of complex context biases that have escaped detection ., Our results also suggest that hydroxymethylation patterns in ES cells are at least in part predictive of hydroxymethylation patterns in an evolutionarily relevant germline context ., Neuron-specific 5hmC sites , which should have no bearing on mutation dynamics in the germline , exhibit rates indistinguishable from matched 5mC sites as predicted ., Conversely , mESC 5hmC sites overlap more frequently than 5mC sites with regions that are enriched for 5hmC during different stages of spermatogenesis ., The results above are consistent with a model where hydroxymethylation has a causal role in generating higher C to G rates at 5hmC sites ., A mutational bias associated with hydroxymethylation might come as a surprise ., Several in vitro studies concluded that 5hmC correctly templates incorporation of G during replication 36–39 , in line with results from structural models that DNA polymerases cannot distinguish 5hmC from 5mC 40 ., Why then , with replication seemingly unaffected , are 5hmC sites associated with increased transversion rates ?, One intriguing lead comes from recent in vitro evidence that 5caC:G pairs stimulate exonuclease activity of polymerase δ and are bound – as strongly as G:T mismatches – by the mismatch repair ( MMR ) complex MutSα , which recognizes post-replicative single-base mismatches 39 ., Thus , base pairs involving oxidized methylcytosines might be mutagenic despite correctly templating G incorporation if they are ( mis- ) recognized as lesions by error-prone DNA repair machinery ., That MMR might be implicated in 5hmC-related mutagenesis is intriguing ., MMR operates immediately after replication when it needs to discriminate the newly replicated from the template strand , thus exhibiting an intrinsic requirement for asymmetry ., In bacteria this requirement is catered for by transient asymmetric methylation 41 ., How such guidance is achieved in eukaryotes remains unclear , but it is interesting to speculate that asymmetries in methylation state might affect and perhaps actively coordinate mismatch repair in eukaryotes ., Intriguingly , examining our cancer data , we discovered strong correlations between %C2G and three components of MMR: MSH2 , MSH6 , and MBD4 ( Figure 5B and Table S3 ) ., MSH2 and MSH6 form the MutSα heterodimer mentioned above , while MBD4 , through an unknown post-transcriptional mechanism , regulates the stability of MSH2 , so that MBD4 depletion reduces the number of MMR-competent MutSα complexes 42 , 43 ., In addition , MBD4 can bind and therefore potentially guide MutSα to methylated and hydroxymethylated CpG sites 44 , 45 ., It is further worth noting that MSH6 – the MutSα protein that makes direct contact with DNA 46 – was recently identified as one of the very few proteins specifically enriched for binding 5hmC 47 ., ( Although a related study did not report preferential binding of MSH6 to 5hmCs 45 , this might be linked to the nature of the probes employed ., Pull-down probes in the former study were made to carry 5hmC via PCR-mediated incorporation of 5hmCTP , an approach expected to lead to 5hmCTP incorporation outside its natural CpG context , thus generating de facto asymmetric sites ., In contrast , the latter study used a synthetic probe that only contained fully symmetrically hydroxymethylated CpGs ( M . Vermeulen , pers . comm . ) ., It therefore seems possible , and consistent with in vitro replication studies , which normally only consider a single modified site , that MSH6 might preferentially associate with asymmetrically hydroxymethylated sites ., This might explain why asymmetrically hydroxymethylated sites suffer from higher mutation rates , as suggested by Figure 5A ) ., Based on these observations we suggest the following model that links MMR , hydroxymethylation and elevated C to G transversion rates:, 1 . ) 5hmC can be further oxidized by the Tet family of enzymes to 5fC and 5caC, 2 . ) During DNA replication , 5caC:G pairing induces exonuclease activity of the replicating DNA polymerase δ and is targeted by MutSα 39 , either incidentally or as part of a regulated process ., 3 . ) MutSα binding triggers MMR ., 4 . ) G to C transversions are introduced by MMR-affiliated translesion synthesis ( TLS ) polymerases ., Alternatively , one might consider a slightly more complex model where mutagenic effects derive from an interaction between the MMR and BER DNA repair pathways: After TDG glycosylase removes 5caC/5fC , the resulting abasic site is hijacked by an MMR-affiliated TLS polymerases , leading to elevated transversion rates ., An analogous scenario has been suggested for the MSH2- and UNG2-dependent generation of C to G transversion by the TLS polymerase Rev1 in the context of somatic hypermutation 48–50 ., This model is attractive because it reconciles recent findings of MutSα binding to 5hmC/5caC with known activity of BER at 5caC and 5fC sites ., Both models predict MutSα binding to be the rate-limiting factor in the generation of C to G transversions ., Detailed biochemical studies will be required to test this hypothesis ., However , it is clear from the analyses presented here that hydroxymethlated and methylated CpGs show differential mutation biases that have left a detectable mark on genome evolution , and we propose differences in DNA repair dynamics as a plausible cause of elevated C to G mutation rates at hydroxymethylated cytosines ., Starting with all cytosine residues in the human reference genome , we confined analysis to cytosines covered by at least 10 reads in the genome-wide bisulfite sequencing of the H1 hESC cell line conducted by Lister et al 16 ( ftp://neomorph . salk . edu/mc/h1_c_basecalls . tar . gz ) , principally to render results more comparable between species and allow detection of lowly ( down to 10% ) hydroxymethylated sites ., We excluded residues that are part of repeats as annotated in UCSC ( hg18 ) and added information on hydroxymethylation status , assayed at base-resolution for the same cell line 15 ., 5hmC , 5mC , and C sites were then defined as described in the main text ., Data on single nucleotide polymorphisms in the human population from the 1000 Genomes Project 51 were obtained using Ensembls biomart facility 52 ( Ensembl Variation 73; Homo sapiens short variation ( GRCh37 . p12 ) ; 1000 Genomes – All; Validated variations only; Minor allele and frequency ) ., The ancestral allelic state was obtained directly from the 1000 Genomes Project ( ftp://ftp . 1000genomes . ebi . ac . uk/vol1/ftp/pilot_data/technical/reference/ancestral_alignments ) ., We combined ( hydroxy ) methylation and polymorphism data after converting all coordinates to hg19 using the liftOver tool 53 , and then confined analysis to nucleotides for which the human ancestral state was unambiguous ( uppercase residues in the 1000 Genomes Project ancestral alignment ) and that were considered assayable by the 1000 Genomes Project ( /vol1/ftp/phase1/analysis_results/supporting/accessible_genome_masks/20120824_strict_mask . bed ) so as to exclude false negative variation calls ., The Ensembl 6-primate alignment ( ftp://ftp . ensembl . org/pub/mnt2/release-75/emf/ensembl-compara/epo_6_primate/ ) was used to reconstruct substitutions along the chimp lineage ., We only considered residues that were cytosines in both human and orang-utan ., The H1 hESC genotype is not the same as the genotype of the human reference genome ., This poses the following problem: Bisulfite sequencing works by protecting 5mC residues but not unmethylated cytosines from being converted to uracil ., Consequently , whenever sequencing reveals the presence of a U/T that maps to a C in the reference , we would infer that we have recovered an unmethylated C . However , we might also be dealing with a site where the H1 genotype deviates from the reference and is in fact T . In this scenario , erroneously assuming the reference genotype to be present would inflate the number of unmethylated cytosines ., This might seem like a minor problem , but can in fact strongly distort downstream evolutionary analysis of unmethylated cytosines , especially when it comes to the analysis of derived allele frequencies ( Figure S2 ) ., To be conservative and enable different downstream analyses , we therefore decided to genotype H1 using available H1-derived short read data ( RNA-seq , Chip-seq , etc . ) from the ENCODE project 54 ., Genotype calls were generated from short read alignment files using samtools mpileup and bcftools 55 , with parameter values depending on the mapping algorithm used for generating a given short read alignment ( see Table S4 for details ) ., Subsequently , we confined analysis to nucleotides covered by at least 20 reads and without a single read suggesting a non-reference genotype ., This strict genotype filtering not only resolves the problem inherent in bisulfite sequencing but also ensures that we are dealing with sites that are homozygous in H1 ., This is important to allow a fair comparison of methylation levels across sites and also facilitates comparison between human and the inbred mouse strains ., As done for human , we start from a list of all cytosines in the mouse genome ( mm9 ) and subsequently remove nucleotides covered by fewer than 10 reads in the bisulfite sequencing study of the E14TG2a mESC line conducted by 14 ., As before , we exclude nucleotides annotated as repeats by UCSC ( based on mm9 ) , added hydroxymethylation 15 and converted coordinates to mm10 using liftOver ., Unlike Lister et al 16 , Stadler and colleagues do not provide binary methylation calls ( methylated/unmethylated ) for the mESC data ., To emulate binary calls in mouse , we examined the distribution of methylation levels for cytosines in a CpG context called methylated/unmethylated in human ., Residues where less than 20% of the reads support methylation are predominantly called unmethylated ( Figure S1 ) and we therefore define mouse residues – which follow a similar bimodal distribution of methylation levels overall – as unmethylated/methylated if less/more than 20% of the reads indicate methylation ., In the absence of extensive genome-wide polymorphism data for wild mice populations , we considered polymorphisms across a collection of laboratory mouse strains sequenced by the Sanger Institute 56 ( available at http://www . sanger . ac . uk/resources/mouse/genomes/ ) , which are derived from three wild sub-species: Mus musculus domesticus , Mus musculus musculus , and Mus musculus castaneus 57 ., We used Mus spretus , a sister taxon to Mus musculus also included in the strain sequencing effort and rat as the outgroup to polarize polymorphisms ., Specifically , based on the mouse-rat ( mm10-rn5 ) pairwise alignment from UCSC we retrieved the corresponding Mus spretus variants and inferred the base ancestral to all Mus musculus strains by parsimony ., We only considered sites where genotype calls were made across all strains and further confined analysis to sites where the genotype was congruent between the mouse reference genome and the 129P2/OlaHsd strain from which the mESC line was derived ., Note that the bisulfite sequencing study of Stadler et al . 14 explicitly took into account the genotype of the mESC line used and only considered cytosines present in the 129P2/OlaHsd strain , so that we did not have to replicate our H1 pipeline and conduct further genotyping ., Local ( ±50 nt ) and regional ( ±500 nt ) GC content around each eligible cytosine as well as the upstream/downstream neighbouring nucleotides were computed from the reference and reconstructed ancestral sequence for mouse and human ., Choice of either ancestral or reference sequence here has no significant impact on the results and we therefore only present data derived from the ancestral sequence ., Chromatin context is strongly associated with mutation rates in cancer genomes 58 and might also affect mutation dynamics in the germline ., At the same time , 5hmC is non-randomly represented across different chromatin states ( see main text ) ., To rule out a confounding effect of chromatin environment on C to G transversion biases , we adopted a popular approach to partition genomic regions into mutually exclusive chromatin states based on the distribution of different histone marks and DNA-binding proteins ., For H1 hESC , we used pre-existing chromatin state calls from the ENCODE project , where information on the genome-wide distribution of 8 histone modifications ( H3K4me1/-me2/-me3 , H3K27ac , H3K9ac , H3K36me3 , H4K20me1 , and H3K27me3 ) and CTCF binding was used to define 15 chromatin states using the ChromHMM algorithm 59 ., For mouse , we collated data on the genome-wide distribution of seven histone marks ( H3K4me1/-me2/-me3 , H3K36me , H3K9me3 , H3K27me3 , H4K20me3 ) in mouse ES cells obtained from two publications 60 , 61 and ran ChromHMM to partition the mouse genome into 14 distinct chromatin states ., The distribution of H3 histones in 60 was used as input ., Coordinates of histone marks w | Introduction, Results, Discussion, Materials and Methods | It has long been known that methylated cytosines deaminate at higher rates than unmodified cytosines and constitute mutational hotspots in mammalian genomes ., The repertoire of naturally occurring cytosine modifications , however , extends beyond 5-methylcytosine to include its oxidation derivatives , notably 5-hydroxymethylcytosine ., The effects of these modifications on sequence evolution are unknown ., Here , we combine base-resolution maps of methyl- and hydroxymethylcytosine in human and mouse with population genomic , divergence and somatic mutation data to show that hydroxymethylated and methylated cytosines show distinct patterns of variation and evolution ., Surprisingly , hydroxymethylated sites are consistently associated with elevated C to G transversion rates at the level of segregating polymorphisms , fixed substitutions , and somatic mutations in tumors ., Controlling for multiple potential confounders , we find derived C to G SNPs to be 1 . 43-fold ( 1 . 22-fold ) more common at hydroxymethylated sites compared to methylated sites in human ( mouse ) ., Increased C to G rates are evident across diverse functional and sequence contexts and , in cancer genomes , correlate with the expression of Tet enzymes and specific components of the mismatch repair pathway ( MSH2 , MSH6 , and MBD4 ) ., Based on these and other observations we suggest that hydroxymethylation is associated with a distinct mutational burden and that the mismatch repair pathway is implicated in causing elevated transversion rates at hydroxymethylated cytosines . | Most cytosines that occur in a CpG context in mammalian genomes are methylated ., Methylation has important functional consequences in the cell but also affects genome evolution ., Notably , methylated cytosines are prone to deaminate and constitute mutational hotspots in mammalian genomes ., Recently , a series of other modifications , derived from the oxidation of methylated cytosines , was shown to exist in various mammalian cell types including embryonic stem cells ., The most abundant of these modifications is 5-hydroxymethylcytosine ., In this work , we ask whether methylated and hydroxymethylated cytosines are subject to the same mutational biases or lead to distinct patterns of genome evolution ., To do so , we examine differences between individuals , between species , and between normal and cancer tissues alongside high-resolution maps of DNA methylation and hydroxymethylation in the human and mouse genomes ., Unexpectedly , we find that hydroxymethylated cytosines are associated with more cytosine to guanine changes in both human and mouse populations , in closely related species , and in the context of somatic evolution in tumors ., Based on multiple lines of evidence , we suggest that the different patterns of sequence evolution at methylated and hydroxymethylated sites are owing to differences in how these sites are handled by the DNA repair machinery . | biochemistry, microevolution, genomics, genome evolution, evolutionary biology, evolutionary processes, genetics, biology and life sciences, dna, computational biology, dna modification, comparative genomics, epigenomics, molecular evolution, dna methylation, epigenetics | null |
journal.pcbi.1005879 | 2,017 | Thalamocortical control of propofol phase-amplitude coupling | Propofol is one of the most popular intravenous anesthetics 1 ., Despite its ubiquity , the neural mechanisms by which it disables consciousness remain poorly understood 2 , 3 ., Characterization of the critical mechanisms of propofol could enable creation of better targeted anesthetics , identification of consciousness-disabling pathways , and more advanced tools to evaluate depth of anesthesia ., This paper presents a novel thalamic sustained alpha rhythm , explores the limits of these thalamic intrinsic oscillations , and describes a thalamocortical origin of propofol phase-amplitude coupling regimes ., The many electroencephalogram ( EEG ) spectral properties of propofol , illustrated in Fig 1 , may provide insight into how it accomplishes its effects ., Beginning at patient loss of consciousness ( LOC ) , its EEG profile consists of a rise in Slow Wave Oscillation power ( SWO , 0 . 1–1 . 5 Hz ) 2 , 4–6 and beta power ( 14–20 Hz ) 7 , 8 that decays to alpha power ( 8–12 Hz ) 3 , 9 , 10 for the duration of the anesthesia ., Recently , the SWO phase and alpha amplitude were found to exhibit phase-amplitude coupling ( PAC ) 6 , 11 , 12: around time of LOC , alpha amplitude is maximum at the SWO trough ( trough-max ) , but as the dose increases and the patient experiences deeper anesthesia , the alpha amplitude switches to maximize at the SWO peak ( peak-max ) ., How propofol controls alpha-SWO PAC is still a mystery , but investigating the neural mechanisms of propofol-induced spectral properties may allow discovery of the key components needed for LOC provided by propofol ., Propofol rhythms are similar in frequency to sleep rhythms associated with the thalamocortical loop , including SWO 2 , 13 , 14 and sleep spindles ( 10–16 Hz ) 9 , 15 ., Mathematical models have played a prominent role in understanding thalamocortical rhythms in both sleep and anesthesia ., There is a strong history of modeling SWO 16–21 , especially in cortex 22–24 , and likewise for thalamic spindle generation 25 , 26 ., More recently , many facets of propofol rhythms have been modeled , including transient propofol beta 8 , burst-suppression 27 , corticothalamic entrainment of alpha 28 , alpha anteriorization 29 , and neuromodulatory impacts on propofol SWO 21 ., However , the fundamental source of propofol alpha , its relationship to anesthetic hyperpolarization , and its relationship to the GABAA potentiation of propofol are all unknown ., Specifically , while we have previously modeled propofol alpha arising from propofol GABAA potentiation and thalamocortical entrainment of this propofol alpha 28 , we did not investigate the mechanisms of its creation ., Additionally , there have been no modeling studies examining either the mechanistic relationship between SWO and alpha under propofol ( including their PAC ) , why this PAC is affected by dose , or the effects of direct hyperpolarization on these dynamics in this system ., Since propofol alpha appears to be sustained over the entire course of anesthesia 11 , 12 , 30 and propofol alpha has been detected in thalamic recordings 31–35 , the relationship of that rhythm to waxing-and-waning spindle oscillations is also unclear ., To better understand how propofol sustained alpha and alpha-SWO PAC are generated , we simulated a thalamic , Hodgkin-Huxley network model under corticothalamic SWO input ., We found that the thalamus is able to generate a sustained alpha rhythm independently of cortical input ., While thalamic intrinsic oscillations are sensitive to both thalamocortical ( TC ) cell H-current maximal conductance ( gH ) and background excitation ( excitation from brainstem and cortex ) 36 , 37 , only under strong propofol GABAA potentiation did we find a persistent , sustained alpha rhythm ., This thalamic sustained alpha may be relayed up to cortex , where it is subsequently seen on the EEG similar to that seen under propofol unconsciousness in the operating room ., Our simulations also showed that the thalamus may control the phase of the alpha coupling to SWO ., Surprisingly , changes to the background excitation alone could control the PAC regime of the system between trough-max and peak-max ., Our analysis of the mechanisms of propofol PAC could help differentiate the roles of thalamus and cortex in propofol phenomena , leading to better understanding of anesthetic action and the components of consciousness ., We began by investigating sustained thalamic alpha; this is the first step in understanding alpha-SWO PAC , in which the alpha rhythm may occur in only parts of the SWO phase ., For all simulations , unless otherwise noted , we considered a 50 thalamocortical cell ( TC ) and 50 reticular cell ( RE ) computational thalamic model for our thalamus ( see Methods ) , inherited from 28 , itself derived from a well-established model 37 ., Under baseline conditions ( no propofol ) , the model thalamus did not produce any intrinsic oscillatory activity , but instead settled into a stable , silent depolarized state after a transient from initial conditions ( Fig 2A ) ., To model low-dose propofol , we doubled both the maximal GABAA conductance , gGABAA , and the decay time constant of GABAA inhibition , τGABAA , throughout the system 8 , 28; to model high-dose propofol , we tripled both gGABAA and τGABAA ., Simulating the same baseline system as above , but under either low-dose or high-dose propofol conditions , we found sustained , persistent alpha firing ( Fig 2B , 2C and 2D ) ., Synaptic effects from propofol can enable sustained alpha firing in the thalamus , and we explore how this comes from intrinsic properties of the thalamus ., The thalamus is known to display a range of behaviors , largely dependent on the excitation level of the system , TC cell gH , and TC cell T-current window interaction: tonic spiking , silent depolarization , hyperpolarized oscillation including spindling and SWO , and silent hyperpolarization 36 , 37 ., This raises the question of whether sustained alpha oscillation can emerge in normal thalamus , especially in oscillations near the alpha frequency range , such as spindles ., Because both the level of background excitation , also known as applied current , and the level of gH are known to be variable and alter the dynamical state of the thalamus 16 , 36 , 37 , we simulated across both of these dimensions to search for regions of sustained alpha oscillations ., Note that our use of background excitation is not meant to model direct inhibition into the system , but rather the sum of hyperpolarizing effects from changes to neuromodulators , second-order neuromodulatory effects , and loss of brainstem excitation ., Since we did not know how weakly or strongly hyperpolarizing the anesthetic was to the system , we modeled the sum of these effects as voltage-invariant tonic charge changes to investigate the entire dimension ., Thus , negative background excitation represents tonic hyperpolarization of the system ., Over the entire physiological range of the gH-background excitation plane , sustained alpha oscillations did not emerge under baseline conditions ( Fig 3 ) ., However , alpha transients in the form of spindles occurred ., These were distinguished from propofol sustained alpha by their waxing and waning nature ., Although propofol decreases gH 38 , these results indicate that decreasing the H-current alone is not sufficient to enable sustained alpha ., Similarly , by varying background excitation to model ascending brainstem neuromodulation , we also found this was not sufficient to enable sustained alpha ., Therefore , sustained alpha is likely not a normal thalamic rhythm , and propofol GABAA potentiation is likely to be responsible for generating sustained alpha ., Our simulations so far suggest that potentiation of GABAA by propofol enables sustained alpha ( Fig 2B , 2C and 2D ) , but we have yet to demonstrate the robustness of these propofol-induced oscillations to changes in gH and background excitation ., To do this , we analyzed the behavior regimes of each simulation under low-dose and high-dose propofol across the physiological gH-background excitation plane in ( Fig 4B and 4C ) ., In baseline simulations in Fig 4A , as background excitation increases , the system shifts from sub-alpha oscillations and transients/spindling into silent depolarization ., When propofol is applied in low-dose and high-dose propofol planes ( Fig 4B and 4C ) , increasing the background excitation enables the system to fire in sustained alpha oscillations ., The sustained alpha oscillations emerge from the same region of the gH-background excitation plane , and even at low-dose propofol it comprised roughly the same area as all other intrinsic oscillations combined ., The size of this sustained alpha firing area roughly doubled when the GABAA potentiation of propofol was tripled ( high-dose , Fig 4C ) compared to doubled ( low-dose , Fig 4B ) ., The region of sustained alpha expanded into states where the background excitation was even higher ., This effect was due to increasing the dynamic range over which excitation and inhibition were balanced ( see next subsection ) ., However , other than this increased propensity for sustained alpha firing , there were no major differences between the sustained alpha oscillations of low-dose versus high-dose propofol ., For this reason , when we discuss mechanisms of the sustained alpha , we do not differentiate between sustained alpha coming from low-dose versus high-dose propofol simulations and may only show high-dose simulations in the interest of brevity ., Note that , as in the baseline simulations , the gH of the system must be below some threshold , ~0 . 024 mS/cm2 , in order for any intrinsic thalamic activity to occur , including sustained alpha ., Therefore , it is necessary but not sufficient to decrease gH below this threshold to produce propofol-induced sustained alpha ., By applying low-dose or high-dose propofol via , doubling or tripling both gGABAA and τGABAA , respectively , we find that the synaptic GABAA effects of propofol enable sustained alpha in a dose-dependent manner ., Having shown that GABAA potentiation by propofol can lead to robust sustained alpha oscillations in the model , we next explored the dynamical mechanisms underlying its production in thalamic networks ., As we compare high-dose propofol simulations to their baseline counterparts in Fig 5 , there are two main facets to how propofol enables sustained alpha in this thalamic system: engaging thalamic spindling dynamics and balancing enhanced inhibition with excitation ., Propofol takes advantage of intrinsic spindling dynamics in producing sustained alpha ., During high-dose propofol , sustained alpha was present where there were either transients/spindles or silent depolarization in the baseline case ( Fig 5 ) ., In the case of transients/spindling , Fig 5A shows both the voltage traces of a terminating spindle at baseline and sustained alpha at high-dose propofol; these simulations correspond to the same point in the gH-background excitation plane and are differentiated only by whether propofol has potentiated GABAA synapses or not ., As in prior modeling work 26 , 37 , the spindle at baseline wanes due to Calcium-based up-regulation of the H-current , as can be seen by the increasing magnitude of current in ( Fig 5D ) ., At baseline , this slow , depolarizing current raises the baseline TC cell membrane potentials above the T-current activation window such that bursting cannot occur ( Fig 5B ) ., In contrast , when GABAA is potentiated by propofol , sustained alpha does not wane , since the RE inhibition consistently hyperpolarizes TC cells into the T-current window , enabling T-current bursts to occur ( Fig 5B ) ., Note that the H-current is more active in the high-dose propofol case than at baseline even though the maximal conductance is the same ( Fig 5D ) , yet the H-current is not strong enough to overcome the enhanced RE inhibition of propofol ( up to a point of gH > 0 . 024 mS/cm2 , see previous subsection ) ., Similarly , in Fig 5H through 5K , propofol enabled sustained alpha oscillations in parameter regimes where , due to the strength of the background excitation under baseline conditions , there would otherwise be silent depolarization ., This new sustained alpha occurred for the same reason: the enhanced inhibition forces TC membrane potentials into the T-current window ., The second major facet of propofol-sustained alpha concerns how propofol changes the balance of excitation and inhibition ., The fundamental cycle of the excitatory inputs to this system ( TC cell H-current and positive background excitation ) balanced against the inhibitory inputs ( propofol GABAA potentiation and negative background excitation ) is illustrated in Fig 5G ., TC cells fire and activate RE cells , which burst and have their inhibition onto TC cells magnified by propofol ., This greater inhibition paradoxically increases the probability of a TC cell burst by hyperpolarizing the TC cell membrane potentials into the T-current window ., Upon entering the T-window for long enough , the T-current de-inactivation state variable ( hT ) builds up , and time until the next burst is decreased by depolarizing effects such as positive background excitation and TC H-current ., The amount of augmenting depolarization affects the frequency of the oscillation , determining if the propofol-infused system oscillates at sub-alpha frequencies like theta or delta , or sustained alpha ., The TC bursts again , and the cycle is reset ., The enhanced inhibition is also the reason for the dose-dependent increase in sustained alpha area on the gH-background excitation plane in Fig 4B and 4C: greater RE inhibition by high-dose vs . low-dose propofol allows high-dose propofol to enable sustained alpha oscillations under a broader range of stronger background excitation ., Increases in propofol from baseline lead to an increase in network frequency up to a maximum ( Fig 6 ) ; the sustained alpha emerges at the peak , and its frequency decreases with further propofol potentiation ., Increases in background excitation increase the thalamic oscillation frequency up to its network frequency maximum , after which the system is too depolarized to oscillate ., These results are consistent with alpha being the maximum thalamic frequency possible via hyperpolarized intrinsic oscillations 28 ., We have shown how propofol enables thalamic sustained alpha , but in order to analyze propofol PAC between alpha and SWO , we constructed a model of corticothalamic SWO UP and DOWN states in the thalamus ., UP states have two effects on the thalamus: cortical firing ( CF ) and a tonic depolarization step , while DOWN states have no cortical firing ( NCF ) and do not include a tonic depolarization step 14 , 20 ., All of our simulations so far have no spiking inputs and can be used to model NCF states ., To model CF , we simulated gH-background excitation behavior planes for all three dose levels under the influence of an AMPA-ergic 12 Hz Poisson spiking process 28 , 39 ., The behavior under low-dose and high-dose conditions are respectively shown in Fig 7A and 7E ., As a result of introducing CF to the thalamic network , the system oscillates throughout more of the parameter space , and slightly less depolarization is needed to elicit thalamic oscillation , but the CF does not significantly alter the sustained alpha oscillation or overall behavior ., Separately , we can model any tonic depolarization steps as increases in background excitation , so the depolarizing step component of the shift from DOWN to UP is accounted for if the system moves to the right on the background excitation axis as in Fig 7A and 7C ., Similarly , a hyperpolarizing step from UP to DOWN is identical to moving the state of the system from the right to the left on the background excitation axis as in Fig 7A and 7C ., Additionally , as dose increases from trough-max to peak-max PAC , we can model the progressive loss of brainstem excitation simultaneously as an overall decrease in background excitation of both UP and DOWN states ., By analyzing the thalamic networks across the gH-background excitation plane , propofol dose , and CF/NCF , we can compare UP states to DOWN states , therefore enabling steady-state SWO modeling of the thalamic network ., During trough-max , we found our thalamic model could couple its alpha oscillation to the trough of the cortical SWO input , driving cortical alpha power trough-max PAC ( Fig 7A through 7D and illustrated in Fig 8 ) ., Under low-dose , trough-max conditions , the thalamic DOWN state is hyperpolarized such that it expresses sustained alpha even though there is no cortical input ., Note that this is a change in hyperpolarization through decreasing background excitation , not a change in gH ., This is illustrated in Fig 7B and 7C , and a simulation is shown in the DOWN states of Fig 7D ., Under a thalamic UP state , however , the incoming CF and increase in background excitation depolarize the TC cells out of their sustained alpha regime , forcing them into their silent depolarized state as shown in Fig 7A ., Thus , the cortex will receive strong alpha input from the thalamus only during its DOWN states , when the thalamus is independent , and this thalamic alpha input will cease during the UP states , allowing the cortex to experience trough-max PAC ., Similarly , during peak-max PAC , we found the thalamic network could be the alpha driver on the cortical SWO peak ( Fig 7E through 7H ) ., As the high-dose thalamus can be more hyperpolarized than the low-dose state , the DOWN state of the thalamus experiences silent hyperpolarization as in Fig 7G ., However , upon receiving CF and depolarization from the cortical UP state , the thalamic UP state exhibits sustained alpha output ( Fig 7E ) ., The cortex receives no thalamic input during DOWN states but receives sustained alpha thalamic input during UP states , producing peak-max PAC observable in the cortex ., Surprisingly , we also found that a change in only thalamic background excitation could enable switching between trough-max and peak-max PAC ., Keeping the state of the system steady on the gH-background excitation plane , increasing GABAA potentiation was insufficient to switch between trough-max and peak-max ., While UP/DOWN alternation in the low-dose propofol thalamus can explain cortical trough-max PAC ( Fig 7A through 7D ) , a decrease in the overall level of background excitation would cause the thalamic UP states to exhibit sustained alpha while the DOWN states exhibit silent hyperpolarization , enabling peak-max PAC even at this lower dose regime ., Thus , peak-max PAC is possible even under a very hyperpolarized thalamus with only moderate GABAA potentiation ., Tripling the GABAA potentiation ( high-dose ) instead of doubling increased the probability of sustained alpha but could not account for a PAC shift under alternating SWO states by itself ., Human EEG studies have shown that , near the onset of and during LOC from propofol , the brain produces an alpha oscillation and a SWO 30 ., These are coupled differently depending on the depth of anesthesia: at the onset or offset of LOC , the alpha rhythm appears in the trough of the SWO ( “trough-max” ) , while at a deeper level of anesthesia , the alpha appears at the peak of the SWO ( “peak-max” ) 12 ., In this study , we have used thalamic Hodgkin-Huxley-type simulations of the effects of propofol on gH , GABAA conductance , and decay time , as well as changes in background excitation , to investigate how the thalamus can control such alpha-SWO PAC ., We found that the hyperpolarized thalamus exhibited a novel thalamic sustained alpha rhythm that only occurs under GABAA potentiation , as in propofol ., Furthermore , depending on thalamic hyperpolarization level , the thalamus expressed alpha during either the corticothalamic SWO trough , producing trough-max output , or during the corticothalamic SWO peak , producing peak-max output ., The first central result is that the thalamus alone is able to produce a sustained alpha rhythm under potentiated GABAA inhibition ( Fig 2 ) ., Thus , we propose a novel thalamic rhythm not observed under normal/awake or sleep conditions , one that requires potentiation of GABAA receptors ., This sustained alpha is fundamentally different than native thalamic spindling due to its lack of waxing-and-waning; it lasted indefinitely , as long as our longest simulations ( 10 seconds ) ., Secondly , this sustained alpha can be produced in the absence of oncoming SWO; this second finding extends the work of 28 by showing the how thalamus alone , independent of patterned cortical input , can robustly produce sustained alpha in the presence of propofol ., The baseline model was able to produce spindling and lower-frequency rhythms , including delta and theta ., However , at no level of depolarization , hyperpolarization , or gH could it produce a sustained alpha rhythm ( Fig 3 ) ., Only propofol potentiation of inhibition , by increasing GABAA conductance and decay time , led to an ongoing alpha-frequency rhythm ., The propofol-induced sustained alpha used the physiological building blocks of the spindling rhythm 25 , 37 ., The mechanism was found to involve the T-current: the added RE cell inhibition led to greater de-inactivation of the T-current in the TC cells , leading to faster and more reliable TC bursts up to a sustained alpha frequency ., With the potentiated inhibition , the sustained alpha appears in regions of parameter space where , in the baseline model without propofol , the network is spindling or silent and depolarized ( Fig 4 ) ., Sustained alpha thus appears between regions of intrinsic oscillations and silent depolarization , created by TC cell membrane potential interactions with the T-current de-inactivation window ., An interesting implication of these results is that direct application of propofol onto hyperpolarized thalamic circuits may be sufficient to produce sustained alpha ., Since the thalamically-generated sustained alpha is expressed in the TC cells , which project to cortex , our models predict that the thalamic alpha could be the source of the coherent frontal alpha observed during anesthetic doses of propofol 11 , 30 ., Our models further suggest that the thalamic alpha is generated independently of propofol-mediated changes to cortical or brainstem circuits , except for the necessary thalamic hyperpolarization ., If this thalamic sustained alpha is consistently relayed to cortex , then we expect the oscillation to be detectable on the EEG ., This prediction could be tested by infusion of propofol into the rodent thalamus alone , along with a hyperpolarizing agent , while recording cortical LFP or EEG and behavior ., Importantly , such experiments may help further characterize the relative role of thalamic alpha in the production of unconsciousness ., Our results also suggest that alpha oscillations are the maximum intrinsic oscillatory frequency achievable by an independent , hyperpolarized thalamus , and this occurs only under GABAA potentiation e . g . by propofol ., The frequency of the ongoing oscillation increased with the conductance and/or decay time of inhibition but not beyond its sustained alpha peak at 300% of the baseline values , the values theoretically associated with high-dose propofol ., This supports earlier computations in 28 , which showed that , as the decay time increased to 300% , the frequency decayed to an asymptotic value ., Increasing the GABAA potentiation to 300% also led to a larger region of parameter space displaying sustained alpha , effectively increasing the probability of the rhythm occurring ., Our study suggests that the H-current effectively acts as a switch between the existence or absence of sustained alpha , and thus potentially as a switch between the conscious and unconscious states ., This is because sustained alpha is possible only if the H-current falls below a threshold level ., Note that the H-current is an intrinsic property of the TC cells and is separable from the level of hyperpolarization of the system ., The thalamic H-current plays a permissive role in our simulations: intrinsic oscillations , including sustained ones , are possible only when gH is adequately small , as in 37 ., Such persistent activity is possible even when this conductance is almost zero , provided there is some depolarizing background excitation ., The effect of propofol on thalamic H-currents is , however , controversial 38 , 40 , 41 ., An H-current switch could be an important component of other anesthesias; sevoflurane has a very similar EEG profile to propofol 42 , and may inhibit the thalamic H-current 43 ., Conversely , this finding has important implications for potentially reversing the anesthetic state , suggesting that agents that increase the H-current in thalamic circuits will shift thalamic dynamics out of the region of sustained alpha oscillation and into the more depolarized awake/relay states ., There are many neuromodulators of the H-current , including dopamine , norepinephrine , and serotonin 44 , and these are widely known to be involved in the sleep-wake cycle; thus their H-current effects may affect consciousness ., Our models suggest that the level of thalamic hyperpolarization is the critical factor determining trough-max versus peak-max coupling as anesthetic dose is increased ., Note that , while propofol GABAA potentiation is necessary for sustained alpha to appear , changing the inhibition level is not sufficient for switching between PAC states ., To achieve peak-max and trough-max coupling , we added another component to our model: cortical spiking , looking separately at thalamic dynamics under the influence of either UP or DOWN states coming from cortex 14 , 20 ., Whereas DOWN state thalamus simply received no cortical input , the UP state thalamus received both a depolarization step , represented by an increase in background excitation , and cortical spiking ., Of note , the UP and DOWN states were modeled the same for all levels of propofol , suggesting that trough-max and peak-max coupling could occur independently of changes to SWO properties ., The trough-max coupling can be produced by introducing propofol GABAA potentiation to a hyperpolarized thalamus during a cortical DOWN state , enabling a sustained alpha ., The corresponding thalamic UP state is too depolarized to express alpha oscillations , so the alpha becomes phase-locked to the SWO trough in the cortical DOWN state ., By contrast , peak-max coupling is possible from hyperpolarizing the thalamus more strongly ( discussed below ) : the input from the corticothalamic UP state helps to compensate for the lower voltage level of the thalamus , enabling alpha oscillations only at the thalamic UP states ., Thus , in peak-max , thalamocortical alpha oscillation occurs only during the peak of the cortical UP state , and there is no major thalamocortical signal during the cortical DOWN state ., Since sustained alpha requires TC cell membrane potential interaction with the voltage-defined T-current window of de-inactivation , the thalamus can switch from trough-max to peak-max merely by becoming more hyperpolarized ( and vice versa ) , even if the cortical activity remains the same ., Increasing thalamic hyperpolarization with propofol dose , an assumption of our model and a critical component of propofol-induced PAC , is supported by several lines of evidence ., In our models , peak-max occurs during decreased , direct background excitation to thalamus that may result from the increased action of propofol on brainstem circuits at high doses 3 , 45 , 46 ., Decreasing the background excitation term may represent potentiation of the potassium leak conductance ( gKL ) from those brainstem neuromodulatory effects , such as decreasing acetylcholine ( ACh ) output , increasing gKL , thus leading to hyperpolarization 47 , 48 ., Recent modeling work has looked at the effects of varying gKL as a proxy for endogenous ACh and histamine ( HA ) changes in thalamic circuits , finding that spindle and SWO oscillations can be generated at certain levels of gKL 21 ., Additionally , physostigmine , a cholinesterase inhibitor , has been found to reverse unconsciousness caused by propofol , ostensibly by enhancing ACh activity 49 ., In the future , we plan to model propofol SWO directly , enabling better understanding of the PAC phenomenon and allowing us to understand the difference between general hyperpolarization and gKL on the thalamic and cortical systems ., This evidence points to the importance of propofol effects in both the thalamus and the brainstem for determining the PAC ., Regardless of hyperpolarization level , there must be thalamic GABAA potentiation from the propofol to enable the creation of sustained alpha oscillations in the thalamus ., Similarly , even if there are sustained alpha oscillations , the thalamus will only exert different PAC regimes if brainstem neuromodulation dynamically hyperpolarizes the system ., We initially thought that both of these changes would have similar effects , allowing us to model both GABAA potentiation and hyperpolarization simultaneously , but this was not the case: the phasic component of GABAA potentiation is critical to creating sustained alpha ( see Fig 5 ) ., The model suggests that trough-max alpha may be more coherent more than peak-max alpha , since trough-max alpha is intrinsically generated by the thalamus , whereas cortical input during peak-max may interfere with the thalamically generated alpha ( Fig 8 ) ., During trough-max coupling , individual thalamic cells synchronize their sustained alpha bursts due to RE cell synchronization ., This greater synchronization in trough-max coupling could increase alpha coherence in cortex and could partially explain the high frontal alpha coherence seen in trough-max 11 , 12 , 42 ., The increased probability of sustained alpha firing in parameter space under peak-max alpha may account for the fact that peak-max alpha is found in more regions across the cortex than trough-max 11 , 12 ., Further experimentation is needed to distinguish between the mechanisms behind such different alpha coherence in cortical circuits during trough-max and peak-max states ., This work has implications for other anesthetics and sedatives ., Dexmedetomidine works in ways similar to sleep pathways by removing excitation to the thalamus and cortex 3 , 10 , 33; experimentally , it produces spindling and SWO at higher doses 50 but not a strong , persistent alpha band that occurs in every SWO cycle ., Our model explains the lack of sustained alpha with dexmedetomidine by the fact that it is not just thalamic hyperpolarization that is necessary but also the change in the time scale of the inhibition produced by GABAA potentiation ., Benzodiazepines , which do change the GABAA time constant of inhibition , can produce beta or alpha oscillations depending on the dose 51 , 52 ., Another GABAA potentiator , sevoflurane also produces coherent frontal alpha and SWO simultaneously 42 ., Recent sevoflurane experiments in rodents have shown that alpha oscillations in the thalamus lead the phase of those in the cortex 53 , suggesting a thalamic source of alpha ., Our model suggests that because sevoflurane induces GABAA potentiation 54 , probable thalamic hyperpolarization , and possible thalamic H-current inhibition 43 , sevoflurane alpha could also be capable of trough-max and peak-max coupling to SWO ., Propofol-induced unconsciousness has been associated with SWO 4 , alpha oscillations 9 , and their PAC 11 ., Our model predicts experimental manipulations that could dissociate alpha and PAC from the SWO component , delineating the role of each of these oscillations in the production of the unconscious state ., Intracranial recordings have shown that humans experience LOC under propofol within seconds of SWO power manifesting 4 , but trough-max PAC has also been found to occur immediatel | Introduction, Results, Discussion, Materials and methods | The anesthetic propofol elicits many different spectral properties on the EEG , including alpha oscillations ( 8–12 Hz ) , Slow Wave Oscillations ( SWO , 0 . 1–1 . 5 Hz ) , and dose-dependent phase-amplitude coupling ( PAC ) between alpha and SWO ., Propofol is known to increase GABAA inhibition and decrease H-current strength , but how it generates these rhythms and their interactions is still unknown ., To investigate both generation of the alpha rhythm and its PAC to SWO , we simulate a Hodgkin-Huxley network model of a hyperpolarized thalamus and corticothalamic inputs ., We find , for the first time , that the model thalamic network is capable of independently generating the sustained alpha seen in propofol , which may then be relayed to cortex and expressed on the EEG ., This dose-dependent sustained alpha critically relies on propofol GABAA potentiation to alter the intrinsic spindling mechanisms of the thalamus ., Furthermore , the H-current conductance and background excitation of these thalamic cells must be within specific ranges to exhibit any intrinsic oscillations , including sustained alpha ., We also find that , under corticothalamic SWO UP and DOWN states , thalamocortical output can exhibit maximum alpha power at either the peak or trough of this SWO; this implies the thalamus may be the source of propofol-induced PAC ., Hyperpolarization level is the main determinant of whether the thalamus exhibits trough-max PAC , which is associated with lower propofol dose , or peak-max PAC , associated with higher dose ., These findings suggest: the thalamus generates a novel rhythm under GABAA potentiation such as under propofol , its hyperpolarization may determine whether a patient experiences trough-max or peak-max PAC , and the thalamus is a critical component of propofol-induced cortical spectral phenomena ., Changes to the thalamus may be a critical part of how propofol accomplishes its effects , including unconsciousness . | Anesthetics make patients lose consciousness , but how they affect brain dynamics is still unknown ., Changes in EEG brainwaves are some of the few noninvasive signals we can use to learn about this ., By analyzing such data , we can develop more targeted anesthetics , expand our knowledge of sleep circuits , and better understand how diseases impact these systems ., The anesthetic propofol is known , among other effects , to increase synaptic inhibition , but it is unclear how these changes induce EEG alpha ( 8–12 Hz ) oscillations and their interaction with slow wave ( 0 . 1–1 . 5 Hz ) oscillations; these signals have been correlated with the state of propofol-infused consciousness ., We simulated a network of thalamic cells to understand the mechanisms generating these signals ., Propofol-potentiated inhibition produced a novel , sustained alpha rhythm in our network ., Changes to the tonic level of depolarization enabled the alpha oscillations to occur at different phases in the slow wave oscillation , as seen clinically with increasing propofol dose ., The thalamus may be critical to propofol-induced alpha oscillations and their coupling to slow wave oscillations ., By understanding the mechanisms generating alpha , we may be able to design experiments to dissociate alpha from slow waves and determine their independent effects on levels of consciousness . | medicine and health sciences, action potentials, depolarization, drugs, membrane potential, anesthetics, brain, brain electrophysiology, electrophysiology, neuroscience, cognitive neuroscience, clinical medicine, brain mapping, bioassays and physiological analysis, pharmacology, electroencephalography, neuroimaging, research and analysis methods, imaging techniques, clinical neurophysiology, consciousness, electrophysiological techniques, thalamus, hyperpolarization, pain management, anatomy, physiology, biology and life sciences, cognitive science, neurophysiology | null |
journal.pcbi.1002856 | 2,013 | Divide and Conquer: Functional Segregation of Synaptic Inputs by Astrocytic Microdomains Could Alleviate Paroxysmal Activity Following Brain Trauma | Post-traumatic epilepsy develops in some but not all head injury cases , depending on the severity of injury and the time elapsed since trauma ., Often there is a latent period between the traumatic event and onset of paroxysmal activity 1 ., Identification of neurological mechanisms underlying this latency to seizures can offer a possibility for therapeutic intervention ., Experimental and modeling studies suggest that this slow transition from normal to paroxysmal activity might depend on homeostatic adjustment of synaptic conductances , connectivity and intrinsic excitability properties 2 , 3 ., Homeostatic synaptic plasticity ( HSP ) likely operates on several spatial and temporal scales 4 ., Chronic synaptic and neuronal inactivity , such as the one that often occurs following trauma , engages glial cells to release tumor necrosis factor alpha ( TNFα ) 5 , 6 , 7 ., This relatively slow process ( global effects in culture are measurable after ∼48 hours of inactivity 6 ) may represent a global “network response” to prolonged inactivity 8 ., The strengthening of inputs from the open eye during monocular deprivation is another slow process that is mediated by TNFα 7 , 8 , 9 ., Early after trauma elevated levels of TNFα are likely to be spatially localized to their glial sources , implying spatial localization of homeostatic synaptic plasticity ., Earlier studies showed that TNFα causes a rapid , p55 receptor mediated insertion of neuronal AMPA receptors 10 , and endocytosis of GABA receptors 6 ., Thus , TNFα could promote epileptogenesis by shifting the excitation-inhibition balance in favor of excitation ., Consistent with this , systemic administration of TNFα 11 and constitutive over-expression of TNFα 12 had pro-epileptic effects ., Seizure incidence was dramatically reduced in knockout mice lacking p55 TNFα receptors 13 , 14; susceptibility to seizures was reduced following systemic pre-injection of TNFα antibodies 15 ., These data suggest that TNFα can promote epileptogenesis 16 ., Given the role of TNFα in HSP 7 , 8 , 9 , the implication is that homeostatic synaptic plasticity can drive the traumatized network toward epileptic activity 3 , 17 , 18 , 19 ., In our previous studies 3 , 18 we showed that trauma-triggered HSP can transform cortical activity from asynchronous spiking ( ∼5 Hz for pyramidal neurons , ∼10 Hz for inhibitory neurons ) to paroxysmal bursting , and we further showed that the pattern of trauma changes the threshold for epileptogenesis 20 ., In those studies we implicitly assumed that HSP represents the action of TNFα which is released in response to chronically low levels of neuronal activity incurred by the traumatic injury ., We also assumed that HSP adjusted synaptic conductances in a manner that depended on the network-global averaging of neuronal activity ., The assumption of global network averaging of neuronal activity is likely to be valid at sufficiently long time after trauma , when levels of TNFα had equilibrated throughout the network ., However , at a short time ( several hours ) after trauma , elevated levels of TNFα are likely to be localized around their glial sources 21 , thus implying spatial localization of HSP and spatially heterogeneous disruption of excitation-inhibition balance that may strongly favor the transition to seizures ., Given the extensive evidence for the dramatic involvement of TNFα in post-traumatic epilepsy 16 , high levels of localized TNFα a short time after brain injury 21 are not consistent with the relatively low incidence of paroxysmal spikes and seizures during that period ., In the present study , we addressed this question by studying the early effects of TNFα mediated HSP hours after trauma ., Homeostatic synaptic plasticity restored the average network firing rate to its pre-traumatic level but transformed asynchronous spiking to paroxysmal bursts ., Thus , we adopted the rate of paroxysmal burst generation ( rather than the network-averaged firing rate ) as a measure of networks propensity to exhibit the transition to seizures ., Paroxysmal bursts of highly correlated population activity in our model resembled interictal epileptiform discharges ( IEDs ) , which are often considered an important diagnostic feature of epileptic seizures 22 , 23 , 24 ., Thus , a higher rate of population bursting in a post-traumatic model network was considered an indicator of stronger propensity to seizures ., With spatially constrained HSP ( “local HSP” ) , representing local synaptic scaling by TNFα , paroxysmal bursts occurred in post-traumatic network at a high rate , with little dependence on the fraction of deafferented neurons ( trauma volume ) ., This was in striking contrast to the gradual dependence of burst rate on trauma volume that characterized the later stage of “global HSP” ., Properties of paroxysmal discharges could be modulated by functional segregation of synaptic inputs into reactive astrocytic microdomains 25 , 26 ., Thus , our modeling studies suggest that some aspects of reactive astrogliosis might alleviate paroxysmal activity early after trauma ., Homeostatic regulation is likely to operate on a localized spatial scale during the early phase of response to trauma , reflecting the local response of glial cells to nearby synaptic activity ., Such a change of scale could in principle affect our earlier conclusions regarding the role of trauma pattern in post-traumatic epileptogenesis 20 ., In particular , in our earlier studies 20 we found that the trauma threshold for the emergence of paroxysmal events in post-traumatic network depended on the pattern of trauma itself ., In those studies , the extent of trauma was parameterized by the fraction of deafferented model neurons , , which we will refer to as the “volume of trauma” ., When burst rate was plotted vs . the trauma volume parameter , focal trauma ( spatially contiguous set of deafferented neurons ) caused lower burst threshold as compared to diffuse trauma ( spatially randomly distributed set of deafferented neurons ) ., Thus , it was important to validate the conclusions of our earlier studies regarding the role of trauma spatial organization in generation of post-traumatic paroxysmal activity ., Here , we assumed that the downregulation of excitatory synaptic conductances in a computational model of a hyperactive pyramidal ( PY ) neuron was determined by the time-averaged firing rate of the postsynaptic neuron , consistent with postsynaptic synaptic scaling 4 ., On the other hand , upregulation of excitatory synaptic input in response to reduced levels of synaptic activity was determined in our model by the time-averaged firing rate of all PY neurons that projected their synapses to a PY neuron under consideration , corresponding to glial scaling of synaptic conductances by TNFα 8 ., Thus , the baseline model is described by “local UP” regulation and “local DOWN” regulation of synaptic conductance ., This model is referred to below as a local HSP model ., To compare with our previous results 20 we also used a global HSP model , in which both pre- and postsynaptic components of homeostatic synaptic scaling were determined by the global , network-averaged , firing rate of model PY neurons ., Thus , the “global HSP” model that was used in our previous studies 20 is described by “global UP” regulation and “global DOWN” regulation of synaptic conductance ., A shift in the spatial scale of HSP rule in the present model would correspond to the transition from the early phase of post-traumatic reorganization ( during which upregulation of synaptic conductance is constrained to glial sources of TNFα ) to the later phase ( of equilibrated levels of TNFα ) ., In some simulations ( e . g . , Figure 1C ) we only changed the spatial scale of presynaptic , upregulating , HSP component from local ( averaged only over those PY neurons that project their synapses to a given neuron , as in “local HSP” or baseline model ) to global ( averaged over all PY neurons in the network , “global UP” model in Figure 1C ) ., Note that the “global UP” model differs from “global HSP” model in that in the former the downregulating postsynaptic HSP component is local ( i . e . , based on the activity of the specific postsynaptic neuron , similar to the baseline model ) ., In other simulations , the downregulating postsynaptic HSP component was removed altogether from the model network , to assess the impact that homeostatic downregulation of synaptic conductance might have on collective activity; this model is referred to as “global UP no DOWN” ., Note again that the “global UP no DOWN” model differs from the “global HSP” model in that in the latter the downregulating component of HSP is present ., In yet other simulations , the set of synaptic inputs in the local HSP scheme was further randomly and evenly partitioned into several groups , for each of which we applied equations that described presynaptic component of HSP ( Materials and Methods ) ., Such partitioning into sub-groups of synaptic inputs was taken to mimic partition of the cortex into astrocytic microdomains 25 , 27 ., Finally , we compared two different patterns of trauma: focal trauma ( in which a spatially contiguous subpopulation of neurons was deafferented , ( e . g . , Figure 1A1 ) ) , and diffuse trauma ( in which deafferentation affected fraction of neurons randomly selected from the entire network ( e . g . , Figure 1A2 ) ) ., In what follows , the “baseline” model network configuration is defined as a network with one microdomain per neuron , local HSP rule , and subject to focal trauma ., Within the “early response” scheme of the HSP based on the activity of presynaptic neurons ( local HSP , see above and Materials and Methods ) , the rate of paroxysmal bursts was significantly higher for focal trauma ( Figure 1A1 ) compared to the spatially diffuse trauma ( Figure 1A2 ) , and this distinction was observed over a wide range of trauma volume parameter values , ., This was generally consistent with earlier studies in which we showed that the spatial pattern of trauma could critically affect the threshold for post-traumatic paroxysmal activity 20 ., In Figures 1A1 , 1A2 , we also plotted the results obtained with the global HSP model ( in which both up and down regulation of synaptic conductance was determined based on the global network-wide average over activities of pyramidal neurons ) , to compare them with the present model , which made use of local HSP ., Although the two models produced the same result qualitatively ( both showed an increase in the rate of paroxysmal activity above some critical threshold of trauma volume parameter , and in both cases the threshold was higher for the diffuse trauma ) , the threshold for paroxysmal activity appeared to be much lower in the case of local HSP rule ., The difference between the two HSP models was also reflected in the dynamics of the network-averaged firing rate of the PY neurons , shown in Figure 1B1 for the case of focal trauma ., In the local HSP model of the focal trauma , and for relatively small trauma volumes ( ) , the network-averaged firing rate of the pyramidal neurons showed a much steeper transition to its post-traumatic target value compared to the more gradual change observed within the global HSP model ( Figure 1B1 , compare solid red lines for local HSP model and dashed red lines for global HSP model ) ., In contrast , after more severe trauma ( ) the approach of the network-averaged firing rate of the model PY neurons toward its post-traumatic target value was more similar for both the local and global HSP scenarios ( Figure 1B1 , black solid and black dashed lines for local and global HSP , respectively ) ., The quantitative differences between the two HSP models , as reflected in the dynamics of the network-averaged firing rate of model pyramidal neurons , were discernible also within the scenario of diffuse trauma ( Figure 1B2 ) ., However , in the diffuse trauma scenario , there was no qualitative difference between the firing rate reorganization dynamics in different HSP models; both local and global HSP models caused gradual recovery of firing rate for relatively small trauma volume ( Figure 1B2 , solid red and dashed red for local and global HSP models , ) and steeper transition in the case of more severe trauma ( Figure 1B2 , solid black and dashed black for local and global HSP models , ) ., Thus , qualitatively , the strongest effect of HSP localization on the post-traumatic reorganization of electrical activity was observed for relatively small volumes of focal trauma ., This is consistent with the results in Figures 1A1 , 1A2 , which show that the primary effect of local HSP is to lower the trauma volume threshold for burst generation and that this effect is more pronounced in the focal trauma scenario ., We next explored the intriguing independence of the paroxysmal burst rate on trauma volume in a model with a local HSP rule ., Several mechanisms in the new model of homeostatic plasticity could have been responsible for this observation ., It could have been a consequence of normalizing the neuronal firing rates , which was implemented by downregulating postsynaptic component of HSP ( based on the firing rate of postsynaptic neuron , as suggested in 4 ) ., Alternatively , the independence of burst rate on the trauma volume could follow from the local scale of the presynaptic component of the HSP , in contrast to the global , network-wide , scale of HSP employed in earlier models of late post-traumatic phase 3 , 18 , 20 ., To test this second possibility , we replaced the local scale of the presynaptic HSP ( for which the averaging of firing rates was performed over the set of those model PY neurons that projected synapses to a given neuron ) with the global scale of the presynaptic HSP ( for which the averaging of firing rates was performed over all model PY neurons ) ., The postsynaptic component of HSP remained local and was determined by the firing rate of the postsynaptic neuron ., As shown in Figure 1C1 , this manipulation on the spatial scale of the presynaptic HSP component did not result in any significant effect on the burst rate – trauma volume relation ., We then tested the possibility that the apparent independence of burst rate on trauma volume was dominated by the postsynaptic downregulating component of HSP ., When the postsynaptic component of HSP was excluded from the model and the global scale scheme was used for the presynaptic component , the linear relation ( in the supra-threshold regime ) between the burst rate and trauma volume was recovered ( Figure 1C2 , open green diamonds ) ., Thus , it appeared that the downregulating postsynaptic component of HSP acted as a permissive factor , either allowing or preventing the burst rate to be modulated by the spatial scale of the presynaptic HSP component ., Earlier studies 3 suggested that post-traumatic paroxysmal activity arises because HSP acts to restore the firing rates of pyramidal neurons to their pre-traumatic value ., Thus , we reasoned that the above dependence of burst rate on trauma volume during early stages of post-traumatic reorganization might , at least partially , be reflected in the firing rates of PY neurons ., Thus , we estimated the dependence of pyramidal firing rate on the neuronal location in the network ., For this analysis , neuronal firing rates were sampled from the center-symmetric strip ( cross section was 5 cells from the center of the strip ) of “neural tissue” , and at each point the firing rates of model PY neurons were averaged over the cross-section of the sampled strip ., With the postsynaptic downregulating component of HSP present , the firing rate of PY neurons ( either deafferented or intact ) was clamped at ∼5 Hz and did not depend on the spatial scale of presynaptic HSP component ( Figures 1D1 and 1E1 ) ., Indeed , the postsynaptic HSP component prevented firing rate of any individual PY neuron to exceed its preset target rate ., The firing rate of the intact neurons never increased and the firing rate of the deafferented neurons reached the target regardless of the presynaptic HSP model ., Therefore the model with global presynaptic scaling was virtually indistinguishable from the baseline model ( with focal trauma and local HSP ) and for both models the firing rate was independent on the trauma volume ( Figures 1E1 ) ., When the global scale presynaptic scheme was combined with exclusion of the postsynaptic downregulating HSP component , the averaged firing rate of intact PY neurons at the traumatized-intact boundary — defined as a set of PY neurons located within one synaptic footprint from the boundary deafferented neuron ( see Materials and Methods ) — displayed strong dependence on the trauma volume parameter ( Figure 1E2 , open green diamonds ) ., In this model , increase of the trauma volume led to an increase in the size of the deafferented ( less active ) PY population and , therefore , required a stronger increase in the firing rate of the intact PY population to keep the overall firing rate constant ., This model allowed an increase because the postsynaptic downregulating HSP component was absent ., Therefore , for intermediate and high trauma volumes ( Figure 1E2 ) , the firing rates of intact PY neurons were higher than those in the baseline model ( with focal trauma and local HSP ) and the firing rates of deafferented neurons were lower ( Figures 1D2 , black squares vs . green diamonds ) ., Surprisingly , for intermediate trauma volumes ( Figure 1C2 , black squares vs . green diamonds ) the relation between the corresponding rates of paroxysmal bursts was opposite to the one observed for neuronal firing rates ( i . e . , for intermediate values of the trauma volume parameter the burst rate in the “global UP no DOWN” model was lower than the burst rate in the baseline model with focal trauma and local HSP ) ., This suggests that the burst rate was limited by the firing rates of deafferented neurons; even when intact neurons fired at higher rate , they only occasionally triggered bursts in the deafferented population ., Thus , although the “firing rate–burst rate” relation hypothesis could qualitatively account for the dependence of burst rate on trauma volume , it failed to explain the quantitative differences between the two HSP scenarios ( local vs . global models ) ., In our previous studies we showed that , in the deafferentation model of cortical trauma , paroxysmal activity is generated by the intact pyramidal neurons located at the boundary between intact and deafferented regions 20 ., Because our previous studies utilized “global HSP” model , it was important to test whether or not the same conclusions would hold as well for networks with “local HSP” scenario ., Figures 2A1 , 2B1 show snapshots of spatial activity in deafferented regions of model networks ( trauma volume parameter ) , for “local HSP” ( Figure 2A1 ) and “global HSP” ( Figure 2B1 ) , and in both scenarios it is seen that paroxysmal activity propagates in a wave-like manner , from intact and into the deafferented part of the network ( in Figures 2A1 , 2B1 image boundaries correspond to the boundary between the intact and deafferented regions of a network , so that only the dynamics in deafferented regions are shown ) ., In Figures 2A2 , 2B2 the spatial spread of activity is further quantified by computing and plotting , for two scenarios , the time-averaged firing rates of the model neurons ., In these plots , the axis are the spatial dimensions of the network grid , and color codes the firing rate of individual neurons , averaged over a long time window ( T\u200a=\u200a50 s ) after the network had reached its post-traumatic steady state ., Destruction of synaptic connectivity between the intact and deafferented parts of the network completely eliminated paroxysmal activity ( Figures 2A3 , 2B3 ) ., This confirms that paroxysmal activity in the deafferented part of the network critically depends on the existence of functional synaptic connectivity with the intact part ., In our previous studies 28 we showed that the rate of post-traumatic paroxysmal bursts may be set not only by the firing rates of intact PY neurons; other determinants of collective activity include the spatial distribution of intact PY neurons and the strengths of their recurrent synapses 28 ., This implies that the spatial scale of synaptic connectivity pattern — the spatial extent to which synaptic connections can be formed , as determined by the size of the synaptic footprint ( see Materials and Methods for details ) — might predispose the traumatized network to become more or less epileptogenic ., Indeed , experimental evidence suggests that , following traumatic brain injury , a network is likely to undergo changes in its anatomical connectivity 17 , 19 , 29 , 30 ., Further , modeling studies showed that these changes in anatomical connectivity could breach the excitation-inhibition balance and generate epileptic-like seizures 2 ., Although reorganization of synaptic connectivity likely occurs on much slower time scales than the ones we study here ( days vs . hours ) , rapid and localized remodeling of synaptic connections was also reported 31 ., We addressed the possible interplay of synaptic connectivity and HSP spatial scales by scaling up the size of synaptic footprint in the model ., The synaptic footprint is the region from and to which a given model neuron could receive or send synaptic connections , and in the model network it was a 10×10 square centered at the neuron under consideration ., As the size of synaptic footprint was scaled ( by scaling the dimensions of the square region from and to which synapses could be received/sent ) , the probability of establishing synaptic contacts inversely depended on the number of potential pre-synaptic partners , as determined by the footprint size , in order to keep the average number of synapses to a given neuron the same , regardless of the footprint size ., This allowed us to avoid conflating the effects of footprint size with an increase in synaptic connectivity ., As Figure 3A1 shows , the rate of paroxysmal burst generation was smaller for larger synaptic footprint sizes ( parameterized as the half-length of square side ) ., This reduction in burst rate was paralleled by an increase in the rate of neuronal firing during the burst , which in turn stemmed from an increase in the number of spikes fired during the burst ( Figures 3B1–3 for sample burst profiles , Figures 3C1 , 2 for quantification of intra-burst spiking activity ) ., By contrast , within the global HSP scheme , manipulations of the synaptic footprint size averaging activity over all model PY neurons had a much weaker effect on the burst-rate trauma-volume relation ( Figure 3A2 ) ., Because burst nucleation in our model required the activity of a certain fraction of intact neurons to be sufficiently correlated ( in order to be able to “ignite” their deafferented postsynaptic partners ) 28 , a reduction in the rate of paroxysmal discharge could signal reduced correlation between burst-triggering intact neurons ., However , correlation between activities of intact neurons on the boundary did not exhibit any remarkable dependence on the size of synaptic footprint ( Figure 3D1 ) ., The correlation between activities of deafferented neurons did grow up with the increasing size of synaptic footprint ( Figure 3D2 ) ., Thus , the reduction in burst rate that was observed for a larger synaptic footprint did not depend on the reduced correlation between burst-igniting neurons ., Another possibility is that a reduced rate of burst generation reflects higher heterogeneity in interconnectedness and synaptic weights for synapses formed among neurons on the boundary between intact and deafferented regions ., Indeed , distributions of HSP scaling factors at PY-PY synapses in post-traumatic steady state ( Figure 3F ) were characterized by larger standard deviations for scenarios with larger footprint sizes ( Figure 3E ) ., These results suggest that heterogeneity of interconnectedness and synaptic conductances at the intact-deafferented boundary could help to alleviate the onset of paroxysmal bursting activity , but this comes at the expense of more intense spiking activity during the burst ., Assuming that the heterogeneity of synaptic organization at the boundary between traumatized and intact regions is likely to be important in post-traumatic epileptogenesis , we sought to identify the physiological mechanisms that might mediate this effect ., Experimental evidence suggests that homeostatic scaling of synaptic conductances might be at least in part mediated by the soluble tumor necrosis factor alpha ( TNFα ) that is believed to be released from astrocytes to compensate for low levels of glutamatergic synaptic activity 8 ., It is well established that astrocytes can sense glutamatergic synaptic activity and respond to it with diverse spatio-temporal patterns of free cytosolic calcium 32 , 33 ., Although a typical astrocyte contacts ∼100 , 000 synapses 34 , in a recent study , the calcium-mediated detection and modulation of synaptic release by astrocytes under physiological conditions was local , with regulation occurring independently along astrocytic processes ( branches ) in groups of 10 s of adjacent synapses 27 ., These findings are consistent with the notion of astrocytic microdomains , with each microdomain responsible for the autonomous regulation of a small cluster of spatially proximal synapses 25 ., Note that microdomains are morphological feature of astrocytes , and thus astrocytic microdomain should not necessarily contact synapses for regulation; however , any synapse that is regulated by an astrocyte belongs , by definition , to unique astrocytic microdomain ., Spatial localization of astrocytic signaling may translate into autonomous regulation of groups of synapses ., Since we consider here the early stage of post-traumatic response ( when a relatively high glial TNFα concentration is likely to be spatially constrained to its release sites ) such autonomous regulation could increase the heterogeneity of synaptic conductances scaled by glia-mediated HSP ., Thus , we hypothesized that functional segregation of synaptic inputs into astrocytic microdomains could help alleviate the rate of paroxysmal discharges in our model networks ., To test this hypothesis , for each model PY neuron the set of all collateral synapses to it ( from PY and IN neurons ) was randomly partitioned into groups ( microdomains ) , such that on average a group of synapses constituted a microdomain ) ., Homeostatic scaling of synaptic conductances ( both glutamatergic and GABAergic ) for each astrocytic microdomain was determined independently by the time-averaged activity of glutamatergic synapses in it ( according to Equations 11 , 12 ) ., Several computational models were developed to describe interactions between astrocytes and synapses 35 , 36 , 37; however , these models linked increased levels of synaptic activity to calcium elevations in astrocytes , and thus cannot explain how low levels of synaptic activity could culminate in astrocytic release of TNFα ., Rather than attempting to develop a detailed mathematical model to describe this process , we assumed here that the ultimate effect of astrocytic microdomain activation is to scale synaptic conductances according to Equations 11 , 12 ., This approximation allowed us to avoid introducing additional complexity associated with biochemical cascades of activation in astrocytes 37 and to focus on the long-term network effects of interactions between neurons and astrocytes ., Figure 4A shows the dependence of paroxysmal discharge rate on the trauma volume , for several scenarios in which synaptic input to each model PY neuron was partitioned into several microdomains ., We considered here scenarios of focal trauma ., For values of trauma volume above a critical threshold , the burst rate still did not depend on the volume of trauma but the plateau value of paroxysmal burst rate now depended on the number of microdomains into which synaptic input set was partitioned ., When plotted vs . the number of microdomains ( for the same trauma volume ) , the burst rate monotonically decreased for larger numbers of microdomains ( Figure 4B ) , and reached an asymptotic level of ∼0 . 3 Hz for microdomains , corresponding to a situation in which each synapse to a PY neuron was associated with an unique microdomain ( each model PY neuron received , on average , 55 synapses from its fellow PY model neurons , and the maximal number of glutamatergic synapses per PY neuron was 75 ) ., For values , some microdomains had zero “synaptic occupancy” ( i . e . , they had no synapses to regulate ) and thus did not take part in homeostatic synaptic plasticity ., Notably , for the same trauma volume , the burst rate that emerged as a result of diffuse trauma also depended on the number of microdomains , but this dependence was much weaker compared to that of the corresponding focal trauma ( Figure 4B , compare closed squares and open circles ) ., Thus , even though diffuse trauma resulted in a lower rate of bursts in the model with one microdomain , for strongly segregated set of synaptic input it resulted in more frequent bursting than did the focal trauma ., The averaged intra-burst firing rates of PY and IN model neurons were higher for stronger segregation of synaptic input into microdomains ( Figure 4C1 ) , as were the average numbers of spikes fired by model neurons during the burst ( Figure 4C2 ) ., Because the firing rates of model PY neurons in our model could influence the outgoing synaptic conductances through the presynaptic part of HSP rule , we also computed the mean HSP scaling factor separately for the set of PY-PY synapses arriving from deafferented model PY neurons and the set of PY-PY synapses arriving from intact model PY neurons ., The value of the HSP scaling factor is directly proportional to the value of synaptic conductance after scaling , and thus could be taken as a measure of how much the outgoing synaptic conductance of the deafferented vs . intact model PY neurons changes as a function of the number of microdomains and the pattern of trauma ., Figure 4C3 shows that the mean HSP scaling factor shows increasing trend as a function of for excitatory input from deafferented neurons , but decreases with larger for excitatory input from intact neurons ., This effect is qualitatively the same for either diffuse or focal trauma scenarios ( Figure 4C3 , closed vs . open symbols ) ., The effect of microdomain partitioning in reducing the rate of paroxysmal bursts was further seen by visual inspection of network activity raster plots ( Figures 4D1 , 2 , 3 ) ., This suggested that the underlying effect of stronger input segregation on paroxysmal burst rate could be similar to that of altered synaptic footprint size ( Figure 3A1 ) – namely , that increased heterogeneity of synaptic input would lead to the decreased burst rate ., Indeed , as Figure 4E1 shows , increasing the synaptic footprint size resulted in the downward offset of burst rate , consistent with the results shown in Figure 3A1 ., Both the mean and the standard deviation of the HSP scaling factor at PY-PY synapses were in general higher for larger synaptic footprint size , for all values of microdomains considered ( Figures 4E2 , 3 ) ., The mean value of HSP scaling factor at PY-PY synapses was nearly independent of , while its standard deviation was generally higher for larger ( Figures 4E2 , 3 ) ., Segregation of synaptic input into microdomains allows a more independent scaling of the inputs from deafferented and intact neurons and thus enhances the correspondence between the firing rate of a given presynaptic neuron and the resulting homeostatic scaling of its downstream synaptic conductance ., As a result , in segregated inputs scenario , synaptic conductance from intact neurons is weaker than synaptic conductance from deafferented neurons ( Figure 4C3 ) ., On the other | Introduction, Results, Discussion, Materials and Methods | Traumatic brain injury often leads to epileptic seizures ., Among other factors , homeostatic synaptic plasticity ( HSP ) mediates posttraumatic epileptogenesis through unbalanced synaptic scaling , partially compensating for the trauma-incurred loss of neural excitability ., HSP is mediated in part by tumor necrosis factor alpha ( TNFα ) , which is released locally from reactive astrocytes early after trauma in response to chronic neuronal inactivity ., During this early period , TNFα is likely to be constrained to its glial sources; however , the contribution of glia-mediated spatially localized HSP to post-traumatic epileptogenesis remains poorly understood ., We used computational model to investigate the reorganization of collective neural activity early after trauma ., Trauma and synaptic scaling transformed asynchronous spiking into paroxysmal discharges ., The rate of paroxysms could be reduced by functional segregation of synaptic input into astrocytic microdomains ., Thus , we propose that trauma-triggered reactive gliosis could exert both beneficial and deleterious effects on neural activity . | Homeostatic plasticity refers to the ability of neurons and neuronal circuitry to adjust their properties in order to maintain physiologically relevant electrical activity notwithstanding perturbations in synaptic input ., Synaptic input is often chronically reduced immediately following brain trauma , and previous studies had suggested that homeostatic synaptic plasticity can aid in the dynamical transition of the traumatized network toward epileptic seizures , a condition known as “post-traumatic epilepsy” ., This form of homeostatic plasticity is mediated by glial cells which release regulatory molecules shortly after trauma ., In this study we used computational modeling to investigate the mechanisms and the implications of glial mediated plasticity early after trauma ., We show that astrocytes ( a subtype of glial cells ) exert both beneficial and deleterious effects on post-traumatic reorganization of neural activity ., This suggests that , in the dysfunctional neuronal network , some aspects of glial-neuronal signaling could alleviate the dynamical transition to pathological activity . | medicine, biochemistry, theoretical biology, neurological disorders, neurology, biology, computational biology, neuroscience | null |
journal.pntd.0001460 | 2,012 | Sero-Epidemiology as a Tool to Screen Populations for Exposure to Mycobacterium ulcerans | Buruli ulcer ( BU ) , a severe necrotizing skin disease , is caused by the environmental pathogen Mycobacterium ulcerans ( M . ulcerans ) ., Globally , it is the third most prevalent mycobacterial disease that affects immunocompetent individuals after tuberculosis and leprosy 1 ., Currently more than 30 countries , mainly in the Tropics and sub-Tropics , are known to report BU cases 2 ., The main countries that are severely affected lie along the Gulf of Guinea and include Ivory-Coast , Ghana , Togo , Benin and Cameroon ., In the highly endemic countries BU is second after tuberculosis as the most prevalent mycobacterial disease 2 , 3 ., However , the global burden of BU is not clear , because efficient and comprehensive reporting systems are lacking in many of the BU endemic countries ., One characteristic of BU is its focal distribution within highly endemic countries ., Most cases occur in remote villages with limited access to the formal health sector , prompting affected people to seek health at traditional healers 4 ., Even today , not all affected communities may be known to the National BU Control Programs ., Therefore reliable tools to detect and monitor the presence of BU in communities are urgently needed ., The disease presentation , which varies between individuals , starts either as a papule , nodule , plaque or edema and if these non-ulcerative early forms are not treated , extensive tissue destruction leads to the formation of large ulcerative lesions with characteristic undermined borders ., Extensive tissue destruction frequently causes disfigurement and long lasting deformities such as loss of limbs and essential organs , like the eye 5 , 6 ., Many features of BU such as the mode of M . ulcerans transmission and risk factors for an infection with the pathogen are not clearly understood ., However , BU is known to occur mainly in children less than 15 years of age and affects people in wetlands and disturbed environments 3 , 7 ., The pathology of BU is primarily associated with the secretion of the cytocidal and immunosuppressive polyketide toxin mycolactone 8 ., Current methods for a laboratory confirmation of clinical BU diagnosis include microscopic detection of acid fast bacilli ( AFB ) , culture of M . ulcerans , histopathology and detection of M . ulcerans DNA by PCR ., Currently , PCR detection of the M . ulcerans specific insertion sequence IS2404 is the gold standard for BU diagnosis 9 ., Yet , PCR requires elaborate infrastructure and expertise and therefore make it out of reach for primary health care facilities in BU endemic low resource countries ., Serology represents a more attractive approach for the development of a simple test format that can be applied to facilities treating BU in low resourced countries ., Unfortunately , various studies have shown that serological tests targeting M . ulcerans antigens are not suitable to differentiate between patients and exposed but healthy individuals as both groups may exhibit serum IgG titers against these antigens 10 , 11 ., However , serology may be a useful tool for monitoring exposure of populations to M . ulcerans , although great antigenic cross reactivity between M . ulcerans , M . tuberculosis , BCG and other environmental mycobacteria complicates this approach ., We previously profiled an immunodominant 18 kDa small heat shock protein ( shsp ) absent from M . tuberculosis and M . bovis as a suitable target antigen for sero-epidemiological studies ., In spite of the presence of sequence homologues in M . leprae and M . avium , Western blot analyses , using a limited number of sera indicated that this protein can be used to distinguish between M . ulcerans exposed and non-exposed populations 10 ., Here we have extended these studies with larger sets of sera ., These sero-epidemiological studies identified a BU index case in a region of Ghana that was regarded , so far , as BU non-endemic ., Ethical clearance for the study was obtained from the institutional review board of the Noguchi Memorial Institute for Medical Research ( Federal-wide Assurance number FWA00001824 ) ., Written informed consent was obtained from all individuals involved in the study ., Parents or guardians provided written consent on behalf of all child participants ., One part of this study was conducted in five districts of the Eastern Region including East-Akim ( EA ) , New-Juaben ( NJ ) , Suhum-Kraboa-Coaltar ( SHC ) , Akwapim South ( AS ) and Akwapim North ( AN ) as well as two districts of the Greater-Accra Region comprising Ga-West ( GW ) and Ga-South ( GS ) ., While EA and NJ report no BU cases and AN only occasionally , the remaining four districts have communities that report BU regularly to the NBUCP ., GW reports the highest number of cases with an annual average number of 100 new cases , followed by AS , GS and the SKC ., This study focused on selected communities within these districts , which are all located along the Densu River ., The other part of this study was carried out in three communities of the Volta Region , namely Torgorme , Gblornu and Kasa ., These communities are situated along the banks of River Volta in the North Tongu district of the Volta Region , which was so far regarded as BU non-endemic ., Torgorme , Gblornu and Kasa have an estimated population of about a 1700 , 350 and 160 , respectively ., Initial community entry was done by first meeting community opinion leaders , which included the disease control officer responsible for the area , the assembly man and chiefs in order to explain the importance of the activity and to solicit their cooperation ., A rough sketch and count of houses along the length and breadth of the community was carried out by walking through the community in order to estimate an approximate number of houses to be surveyed ., The area was then divided into two blocks , with one research team being responsible for one block ., Each habitable structure within a block was then numbered serially ., A house to house survey was carried out and interviews involving the head of a house were done ., A data collection chart was used to collect information on the number of people in the house , healed and active BU cases and if active cases were found , samples were collected for confirmation of BU ., Collected data of the BU patients included age and sex , when the disease was contracted and GPS coordinates of their houses ., Two milliliters of blood were collected into vacutainer tubes ( BD ) from participants of ten different villages within a 5 km radius along the Densu River; six and four of the communities were confirmed as BU endemic and non-endemic , respectively , using active search and mapping activities as described above ., The endemic communities were: Kojo-Ashong and Otuaplem in the GW , Kwame Anum and Ayitey Kortor in the GS , Sakyikrom and Tetteh Kofi in the AS district , respectively ., The non-endemic communities were Obuotumpan and Abotanso in the NJ and Abesim Yeboah and Ntabea in the EA district ., The study participants aged between 5 and 90 years recruited from these communities were individuals with no history of BU ., 188 participants were from the non-endemic villages ( 94 each females and males ) ; age range 5–84 years , arithmetic mean of 28 . 6 years , median 19 years and mode of 15 ., 294 participants were from endemic villages ( 139 and 155 were male and females , respectively ) ; age range 5–90 years , mean age of 26 . 8 years , median 20 and mode 12 years ., In addition , whole blood samples were also collected from 99 community members in three villages along the Volta River in the Volta Region which has so far been considered one of the non-endemic regions in Ghana ., The three communities Torgorme , Gblornu and Kasa were selected as having never reported leprosy in the past five years according to data of the North Tongu District Directorate of Health Services ., Blood samples were transported immediately at ambient temperature to the laboratory for separation of serum by centrifugation at 2 , 000 g for 10 mins to remove the clot ., Sera were stored at −80°C until analysis ., 25 µg of recombinant M . ulcerans 18 kDa shsp protein was separated on NuPAGE® Novex 4–12% Bis-Tris ZOOM™ Gels , 1 . 0 mm IPG well ( Invitrogen ) using NuPAGE ® MES SDS Running Buffer ( Invitrogen ) under reducing conditions and transferred to nitrocellulose membranes ., Membranes were blocked with 5% skim milk in phosphate-buffered saline ( PBS ) , 0 . 1%Tween 20 ( PBS-T ) and cut into strips ., Protein strips were incubated with serum samples at a 1∶500 dilution in PBS-T for 1 . 5 hrs ., Strips were washed with 0 . 3 M PBS , 1% Tween 20 and incubated with alkaline phosphatase-conjugated AffiniPure F ( ab′ ) 2 fragment goat anti-human immunoglobulin G ( IgG , Milian ) ., Nitro blue tetrazolium ( NBT ) and 5-bromo-4-chloro-3-indolyl phosphate ( BCIP ) ( BioRad ) were used for color development ., 96-well Nunc-Immuno Maxisorp plates ( Thermo Scientific ) were coated with 0 . 5 µg recombinant 18 kDa shsp per well in 100 µl PBS ., Plates were incubated at 4°C overnight ., Plates were washed with dH2O , 2 . 5% Tween 20 ( dH2O-T ) and blocked for 1 h with 200 µl blocking buffer ( 5% skim milk in PBS ) at 37°C ., Serial 2-fold dilutions of serum from 1∶100 to 1∶12800 in 50 µl blocking buffer per well were incubated for 1 . 5 hrs at 37°C ., The wells were washed with dH2O-T ., 50 µl of 1∶6000 diluted goat anti-human IgG ( γ-chain specific ) coupled to horseradish Peroxidase ( HRP , SouthernBiotech ) was added to each well and incubated for 1 h at room temperature ., After the last washing step with dH2O-T , 100 µl TMB Microwell Peroxidase Substrate ( KPL ) was added ., The reaction was stopped after 5 min ., The absorbance was measured using an ELISA plate reader ( Sunrise , Tecan ) at 450 nm ., Each ELISA plate contained two-fold dilutions of a negative control comprising a pool of 5 negative sera from people living in BU non-endemic communities in Ghana and a positive control consisting of 5 medium positive sera from people living in BU endemic areas ., The cut-off value for positivity was considered to be the mean optical density ( OD ) of negative and positive control at a 1∶100 serum pool dilution ., Statistically , data were analyzed using GraphPad Prism version 5 . 0 ( GraphPad Software , San Diego California USA ) ., The nonparametric Kruskal-Wallis test with Dunns post-test was used to compare OD values for the different groups ., Sampling was done from aquatic environments and from communities ., Water , insects , fish , snails , dominant vegetation ( both dead and living ) and soil were collected randomly from the ground and edges of rivers at various locations in both endemic and non-endemic communities ., Soil , vegetation and animal droppings were collected from various locations within both endemic and non-endemic communities ., All collected samples were transported on the same day to the laboratory , stored at 4°C and analyzed within a week of collection ., DNA was extracted from about 200 mg portions of all the environmental samples using the FastDNA Spin kit for soil ( MP Biomedical ) according to the manufacturers instruction ., For insect samples additionally glass beads were added to the lysing matrix and the breaking step with the Fast Prep instrument was substituted by heating specimens at 95°C for twenty minutes followed by vortexing full speed for two minutes ., The extracted DNA was stored at −20°C until analysis by real-time PCR ., TaqMan real-time PCR was performed using primers and procedures as previously described with some modifications in reaction conditions 12 ., The primers and TaqMan MGB probes detecting IS2404 , IS2606 and the ketoreductase ( KR ) domain were obtained from Applied Biosystems ( Foster City , CA , USA ) ., IS2404 real-time PCR mixtures contained 1X Qiagen master mix ( containing HotstarTaq plus DNA polymerase , dNTP mix and PCR buffer ) 1 µl of extracted template DNA , 0 . 5 µM concentrations of each primer and 0 . 2 µM probe , 1× TaqMan exogenous internal positive control ( IPC ) and probe reagents ( Applied Biosystems ) , in a total volume of 20 µl ., Amplification and detection were performed with the Rotor-Gene Q ( Qiagen ) using the following program: 1 cycle of 95°C for 5 min , 40 cycles of 95°C for 15 s and 60°C for 15 s ., Each PCR run contained 2 non-template controls and an IS2404 positive control ., Analysis for IS2606 and KR was in a multiplex PCR using KR and IS2606 probes with FAM and VIC fluorescence labels respectively and reaction conditions as above ., Initial BU survey results were entered in Microsoft Access and exported for integration using Quantum Geographic Information System ( GIS ) for analyses ., Google Earth aerial images of communities were obtained , geo-referenced and linked to ground contours , features and other characteristics ., The prevalence of BU was calculated by counting all individuals in the community with a classical BU scar , together with those with laboratory confirmed active disease , divided by the total number of persons examined within a community ., The rate was expressed as a percentage ., We determined M . ulcerans 18 kDa shsp-specific serum IgG titers in 482 sera from people living in the BU endemic Densu River Valley in the Gar and Eastern Region , 99 sera from people living in the BU non-endemic Volta Region and 20 sera from European controls without travel history to Africa ( figure 1A ) ., Based on the defined ELISA OD cut-off values , a sero-positivity rate of 32% was observed for the sera from the Densu River Valley ., The sero-positivity rate of people living in the Volta Region ( 12% ) , as well as the mean ELISA readouts obtained with their sera were significantly lower ( p<0 . 001 ) ., None of the sera from European controls exhibited a significant titer ( figure 1A ) ., Sero-positive individuals from the Volta region were re-visited and interviewed ., It was determined that all of them have lived entirely or at least for most of their life in their home communities in the Volta Region ., One of the sero-positive participants from the village Torgorme reported at the interview to have a non-healing chronic wound on the leg ( figure 1B ) ., The wound was clinically diagnosed by an experienced physician as BU and clinical diagnosis was laboratory reconfirmed by positive IS2404 PCR of swab specimens ., The serum of this reconfirmed BU patient had the highest anti-18 kDa shsp-specific serum IgG titer of all participants from the Volta Region tested ( figure 1A ) ., Following the identification of this index case , the health directorate of the Volta Region sent us specimens from eleven other individuals with suspected BU lesions ., Six of these , were reconfirmed as IS2404 PCR-positive BU by our laboratory at the Noguchi Memorial Institute for Medical Research , which is one of the BU reference laboratories in Ghana ., While sera of two laboratory confirmed BU patients contained anti-18 kDa shsp IgG , four patients were sero-negative ., Active case search surveys were performed to determine the prevalence of BU along the Densu River ( figure 2 ) ., The average prevalence of BU in endemic communities with 3 km buffer was 3 . 4% ., While in some communities upstream no BU cases were found , the disease burden increases as the River runs downstream ( figure 2A ) ., Of the ten communities included in the sero-epidemiological study , four ( Ntabea , Abesim-Yeboah , Obotanso and Obuotupan ) were confirmed as non-endemic , as both the passive surveillance by the National BU control program and our active case search identified neither healed nor active cases ., The total prevalence rate , including both healed and active cases , of the six endemic communities ranged from 1% to 19% with Tetteh Kofi , Otuaplem and Sode having the highest rates ( 4 . 8% , 14 . 9% and 19 . 1% , respectively ) ., The prevalence of active cases ranged from <1% to 2 . 4% , with Sode also having the highest active case prevalence rate ., When 18 kDa shsp-specific serum IgG titers of 295 sera from BU endemic and of 187 sera from non-endemic communities were analyzed by ELISA ( figure 1A ) , comparable sero-positivity rates ( 33% versus 31% , respectively ) were found ., ELISA results were reconfirmed by Western blot analysis with a randomly chosen subset of sera ., There was good agreement between Western blot band intensities and ELISA titers with a few discrepancies related to a higher sensitivity of the ELISA method ( data not shown ) ., Sero-responders were found in all age groups ( >5 years ) tested , but sero-negative individuals dominated throughout ( figure 3 ) ., 211 environmental samples were collected randomly from both aquatic and dry land environs ., The sampled BU endemic communities included Kojo-Ashong ( KA ) , Sode ( SD ) , Amasaman and surrounding hamlets ( AS ) , and Kudeha and surrounding hamlets ( KD ) located in the GW and GS districts ., Samples from non-endemic communities were collected in Abesim-Yeboah ( AY ) , Obuotumpan ( OB ) and Ntabea ( NB ) located in the EA and NJ districts further up-stream of the Densu River ( figure 2A ) ., M . ulcerans DNA in an environmental sample was confirmed by the presence of all three tested loci ( IS2404 , IS2606 and KR ) as revealed by positive results with all three PCR tests performed ., In all , 19/211 ( 9 . 0% ) of the samples tested were positive , including 5/19 aquatic snails , 5/28 sand samples collected from the communities , 4/30 samples from river water and river bed sand , 2/30 samples from aquatic vegetation , 1/6 sand samples collected from farms , 1/12 aquatic insects and 1/1 millipedes ., As shown in Table 1 the average positivity rates for samples from endemic communities were 13 . 4% ( 7 . 7% , 13 . 3% , 14 . 3% and 18 . 5% in KA , AS , SD and KD , respectively ) and 6 . 2% for samples from non-endemic communities ( 26 . 0% , 2% and 1 . 8% for NB , OB and AY , respectively ) ., Broad antigenic cross-reactivity between mycobacterial species represents a major challenge for the development of a serological test that is specific and sensitive enough to monitor immune responses against M . ulcerans in populations where exposure to M . tuberculosis and BCG vaccination is common ., In our earlier work , we have identified the M . ulcerans 18 kDa shsp as an immunodominant antigen , which has no homologues in M . tuberculosis and M . bovis 10 ., However , interspecies cross-reactivity of this protein with an 18 kDa protein of M . leprae as well as a 20 kDa protein of M . chelonae was detected ., In the same study we evaluated the use of measuring anti-18 kDa shsp IgG titers for assessing the exposure of a population to M . ulcerans on the basis of a limited number of BU patients , household contacts and people living in areas where BU is not endemic 10 ., Since sera from inhabitants of BU non-endemic regions showed largely no reactivity with the 18 kDa protein of M . ulcerans , immune responses against environmental mycobacteria , such as M . chelonae , do not seem to compromise the developed serological test for M . ulcerans exposure ., Here we have extended our previous analysis by comparing sera from areas of Ghana , which rarely report leprosy cases , but differ in their reported BU endemicity ., In Ghana , a national case search performed in 1999 yielded a crude national BU prevalence rate of 20 . 7/100 , 000 and hence demonstrated that BU is the second most common mycobacteriosis in the country after tuberculosis 13 ., In this study diagnosis of both active and healed lesions was based solely on clinical grounds without any microbiological confirmation ., Since the creation of the national control program , 32 of the 166 nation-wide districts continuously report BU ., Through this passive surveillance system , over 11 , 000 cases have been reported between 1993 and 2006 ( http://www . who . int/mediacentre/factsheets/fs199/en/ ) from mainly six of the ten regions of Ghana ., No BU cases have been reported from the Volta , Northern , Upper East and West regions , giving the impression that those four regions do not harbor BU cases and therefore are non-endemic ., However , in our analysis of sera from the Volta Region , a relatively small , but significant number of serum samples contained anti-18 kDa shsp IgG ., Follow-up visits and interviews revealed that one of the sero-positive individuals had a chronic wound which was subsequently laboratory confirmed as BU 14 ., After identification of this index case , additional laboratory confirmed BU cases were found by active case search in the Volta Region ., In our previous analyses 10 , only part of the sera from laboratory reconfirmed BU patients were tested postitive for anti-18 kDa shsp IgG ., In accordance with these findings , not all of the BU patient sera from the Volta region were sero-positive ., These data clearly show that anti-18 kDa shsp IgG titers are no indication for active disease ., A large epidemiological survey is now required to determine the prevalence of BU over the entire Volta Region ., Until today no serological test allows for a distinction of BU patients and healthy individuals , which are exposed to M . ulcerans ., However , our results demonstrate that sero-epidemiological studies can be used to complement active case search in regions , where data about the BU prevalence are lacking ., Future longitudinal sero-epidemiological studies are planned in order to monitor the exposure of certain populations to M . ulcerans over a longer period of time ., At this stage we cannot conclude how timing and frequency of exposure influences antibody titers against the pathogen ., The prevalence of 18 kDa shsp sero-positive individuals within populations along the Densu River was >30% ., This confirms our earlier conclusion that a large proportion of healthy individuals living in endemic communities who have responded immunologically to M . ulcerans exposure do not develop overt disease ., While the percentage of 18 kDa shsp sero-positive individuals was higher compared to that found using a Burulin skin test in healthy controls , it is comparable to that obtained for the serologic response to M . ulcerans culture filtrate 15 , 16 ., Diverse outcome of infection with the causative agents of the main mycobacterial diseases such as tuberculosis seems to be a common feature of their natural history ., Not all exposed individuals show immunological evidence of infection and of those who get infected by M . tuberculosis estimations indicate that only 10% will ever develop overt disease 17 , 18 ., Manifestation of the disease ranges from self-limited pulmonary infection to localized extra-pulmonary infection and disseminated disease 19 ., Factors accounting for the diversity in outcomes are not entirely known , but may relate to both host and pathogen factors ., Even though clinical M . ulcerans isolates from Africa are clonally related and genetically largely monomorphic 20–27 , differences in virulence among African M . ulcerans strains cannot be ruled out completely ., Hence , the percentage of M . ulcerans infected individuals who proceed to develop BU remains to be established ., BU is known to develop in all age groups with a nearly equal gender distribution but most cases occur in children 15 years of age or younger 28 ., In our study we found anti-18 kDa shsp sero-responders in all age groups ( >5 years ) analyzed ., A future cohort study with infants could provide important insight , at which age these immune responses start to emerge ., Both in endemic and non-endemic villages of the Densu River Valley we found M . ulcerans PCR positive environmental samples ., This is indicative for the presence of M . ulcerans or of closely related environmental bacteria all along the Densu River ., Our findings are consistent with earlier findings of Williamson et al . 29 in the same region ., Since the mode of M . ulcerans transmission and risk factors for the exposure to the pathogen are still not entirely elucidated , it is not clear , whether the types of environmental samples that were PCR positive have direct relevance for infection with M . ulcerans ., Hence methods for the routine isolation and characterization of M . ulcerans from the environment need to be developed ., Hypotheses on risk factors and the mode of infection with M . ulcerans include contamination of wounds from an environmental reservoir , inhalation of vaporized contaminated water and inoculation by insects 30–32 ., Our molecular epidemiological studies have recently demonstrated a focal transmission pattern for M . ulcerans 27 ., This may help to explain one of the mysteries of BU transmission , the close proximity of endemic and non-endemic villages ., As indicated in figure 2 , while M . ulcerans is endemic in some villages within the Suhum-Kraboa-Coaltar district , through active case search we did not find any case ( both healed and active ) in neighboring districts located at the upper part of the river , such as East-Akim and the New-Juaben ., In contrast , communities of the four districts , which are situated downstream ( Akwapim South , Akwapim North , Ga-West and Ga-South ) regularly report BU cases ., BU endemic and non-endemic communities along the Densu river differ in terms of their vegetation ., Upstream , within the wet semi-equatorial zone , the vegetation is predominantly moist semi-deciduous rain forest , which gradually changes downstream into a short stretch of Guinea Savannah around Nsawam and ends with coastal scrub and savannah grassland in the Ga districts ., In addition , there is a variation in the features of the Densu River , which takes its source from the Atewa Forest Range near Kibi and flows for 116 km into the Weija Water Reservoir before entering the Gulf of Guinea through the Densu Delta Ramsar site ., While upstream the river flows fast , has clear water and the river bed consists of rocky stones , downstream the river flows sluggishly , has a muddy river bed , and the water is turbid ., We did not find significant differences in anti-18 kDa shsp IgG seropositivity rate or titers between people living in communities in the Densu River Valley that were classified based on active case search as BU endemic or non-endemic ., These findings could imply at least one of the following:, 1 ) people in the non-endemic communities in the upper Densu River Valley may be exposed to M . ulcerans lineages with low virulence;, 2 ) currently unknown host genetic , behavioral or socio-economic factors trigger the development of subclinical M . ulcerans infection to clinical disease;, 3 ) in the non-endemic communities 18 kDa shsp binding antibodies are triggered by subclinical infections with environmental mycobacteria harboring antigens that are cross-reactive with the 18 kDa shsp 10 . | Introduction, Materials and Methods, Results, Discussion | Previous analyses of sera from a limited number of Ghanaian Buruli ulcer ( BU ) patients , their household contacts , individuals living in BU non-endemic regions as well as European controls have indicated that antibody responses to the M . ulcerans 18 kDa small heat shock protein ( shsp ) reflect exposure to this pathogen ., Here , we have investigated to what extent inhabitants of regions in Ghana regarded as non-endemic for BU develop anti-18 kDa shsp antibody titers ., For this purpose we determined anti-18 kDa shsp IgG titers in sera collected from healthy inhabitants of the BU endemic Densu River Valley and the Volta Region , which was so far regarded as BU non-endemic ., Significantly more sera from the Densu River Valley contained anti-18 kDa shsp IgG ( 32% versus 12% , respectively ) ., However , some sera from the Volta Region also showed high titers ., When interviewing these sero-responders , it was revealed that the person with the highest titer had a chronic wound , which was clinically diagnosed and laboratory reconfirmed as active BU ., After identification of this BU index case , further BU cases were clinically diagnosed by the Volta Region local health authorities and laboratory reconfirmed ., Interestingly , there was neither a difference in sero-prevalence nor in IS2404 PCR positivity of environmental samples between BU endemic and non-endemic communities located in the Densu River Valley ., These data indicate that the intensity of exposure to M . ulcerans in endemic and non-endemic communities along the Densu River is comparable and that currently unknown host and/or pathogen factors may determine how frequently exposure is leading to clinical disease ., While even high serum titers of anti-18 kDa shsp IgG do not indicate active disease , sero-epidemiological studies can be used to identify new BU endemic areas . | Sero-epidemiological analyses revealed that a higher proportion of sera from individuals living in the Buruli ulcer ( BU ) endemic Densu River Valley of Ghana contain Mycobacterium ulcerans 18 kDa small heat shock protein ( shsp ) -specific IgG than sera from inhabitants of the Volta Region , which was regarded so far as BU non-endemic ., However , follow-up studies in the Volta Region showed that the individual with the highest anti-18 kDa shsp-specific serum IgG titer of all participants from the Volta Region had a BU lesion ., Identification of more BU patients in the Volta Region by subsequent active case search demonstrated that sero-epidemiology can help identify low endemicity areas ., Endemic and non-endemic communities along the Densu River Valley differed neither in sero-prevalence nor in positivity of environmental samples in PCR targeting M . ulcerans genomic and plasmid DNA sequences ., A lower risk of developing M . ulcerans disease in the non-endemic communities may either be related to host factors or a lower virulence of local M . ulcerans strains . | immunology, biology, microbiology, population biology | null |
journal.pcbi.1002884 | 2,013 | Positional Bias of MHC Class I Restricted T-Cell Epitopes in Viral Antigens Is Likely due to a Bias in Conservation | The immune system rapidly detects virus-infected cells through cell-surface presentation of viral peptides to T-cells despite the fact that the half-lives of source proteins are typically orders of magnitude longer than the response time ( i . e . days vs . minutes ) 1 , 2 ., To explain this paradox , the DRiP hypothesis proposed that defective ribosomal products , rapidly degraded forms of standard gene products , are a major source of peptides for MHC class I processing pathway 3 ., Although the DRiPs hypothesis is well into its teens , surprisingly little is known about the nature of the DRiPs 4–6 ., This is likely due to the low abundance of DRiPs relative to the abundance of folded proteins , which poses a significant challenge to current biochemical/molecular technologies ., A central question is what mechanisms dominate in the production of DRiPs 5 , 6 ., One set of mechanisms that leads to the generation of partial protein products is downstream initiation and premature terminations when translating proteins from mRNAs 7 ., If such errors are involved in generation of DRiPs , a bias in sampling of regions of mRNAs may result ., This sampling bias in turn may influence from which regions of proteins epitopes are more often detected , resulting in positional bias of epitopes ., For example , a dominance of downstream initiation would result in more epitopes detected from the C-terminal ends; conversely , a dominance of premature termination would result in more epitopes detected from the N-terminal regions ., To overcome the limitation of the current experimental approaches , we have investigated if a data-driven approach can provide insights into the nature of DRiPs ., Namely , the availability of a large repository of immune epitopes stored at the Immune Epitope Database ( IEDB ) 8 makes it possible to indirectly characterize the population of DRiPs by measuring positional bias of viral epitopes ., Because viruses exploit the translational machinery of the host to synthesize their proteins and are thus relevant in the context of the DRiPs hypothesis , we retrieved all MHC-I restricted T-cell epitopes of viruses for which reference proteomes are available ., Top 20 viruses based on number of tested peptides are shown in Table, 1 . For each virus , total number of unique peptides tested , peptides with positive T-cell outcomes ( i . e . epitopes ) , and those with negative outcomes are shown ., In the table , vaccinia virus contributed the greatest number of tested peptides , followed by hepatitis C Virus , lymphocytic choriomeningitis virus , human herpesvirus 5/4 and influenza A virus ., After determining positions of viral peptides in their reference antigens and calculating normalized positions , we constructed distributions of normalized positions for positives ( i . e . epitopes ) and negatives as shown in Fig . 1A and 1B , respectively ., Each distribution used 5 bins of equal intervals covering a range 0 , 1 ., Distributions plotted using 10 bins showed similar patterns ( data not shown ) ., The distributions correspond to probability mass functions , p ( x|positive ) and p ( x|negative ) , where ‘x’ represents a binned normalized position ( described further in Methods section ) ., To show positional bias of epitopes , a probability ratio plot ( i . e . p ( x|positive ) /p ( x|negative ) , see Methods for details ) is shown in Fig . 1C ., By dividing the distribution for positive peptides by the distribution of negative peptides , we implicitly take into account study biases that are otherwise difficult to capture ., If positional bias is absent , the corresponding probability ratio plot would show a horizontal line at a probability ratio of, 1 . If positional bias is present , regions with lower likelihoods of finding epitopes would show values less than, 1 . Indeed , in Fig . 1C , the probability ratio plot shows under representation of epitopes at N- and C-termini ., Furthermore , probability ratio plots generated after cumulative exclusion of data from vaccinia virus and Hepatitis C virus , which contributed most in terms of data , resulted in maintaining the overall pattern ( supplementary Fig . S1 ) The positional bias of epitopes observed is supported by results of statistical analyses on the corresponding contingency table ( supplementary Table S1 ) ., Using a binomial test for each bin , it was determined whether counts of positives significantly deviate from the expected fraction of positives ( i . e . ( all positives ) / ( all positives+all negatives ) =\u200a2394/12974 ) ., Out of 5 bins , 4 deviated from the expected ( two sided; p-value<0 . 05 ) ., A possible source of positional bias of epitopes is unequal distributions of amino acids spanning the length of antigens ., For instance , the positional bias observed may have been due to hydrophobic residues being preferentially found in middle regions rather than at N- and C-termini ., Such unevenness would mean that MHC alleles binding peptides with hydrophobic anchor residues would impose positional bias of epitopes ., To rule out this possibility , we determined positional bias of predicted binding peptides of viruses to MHC molecules ., Our prediction strategy uses recently developed peptide:MHC-I binding algorithms , which have achieved high accuracies in benchmarks 9 , 10 ., Using 9-mer peptides generated from a set viral proteins , we made binding predictions to human MHC-I molecules ( HLA ) ., As a note , HLA molecules can be grouped into 12 HLA supertypes based on their known overlap of binding specificities 11 ., SMMPMBEC was used to make binding predictions 12 ., Corresponding member alleles for each supertype used for predictions are provided in the supplementary material ( Table S2 ) ., After grouping predictions based on supertypes , for each supertype , we generated distributions of normalized positions for predicted ‘binders’ ( i . e . those peptides with predicted binding affinities <500 nM ) and ‘non-binders’ ( i . e . those with affinities >500 nM ) ., Probability ratio plots derived from the distributions for 12 HLA supertypes are shown in Fig ., 2 . The lack of divergence from a horizontal line at 1 . 0 indicates absence of positional bias ., In the figure , some of the supertypes show slight positional bias ., Specifically , supertype A01 favors the middle region of antigens , whereas A03 favors the C-terminus ., A combined plot generated from weighted averaging of probability ratio plots from all supertypes based on frequencies of known MHC restriction data showed absence of positional bias ( supplementary Fig . S2 ) ., To ensure that these observations are not due to the choice of predictive method , we used a different predictive method NetMHC 13 , which is another accurate predictive method , and made the same observations ., In addition , scatter plots of predicted binding affinities for the tested peptides and their normalized positions showed no systematic effects ( supplementary Fig . S3 ) ., The presented results largely reflect results reported in 14 , where mammalian , bacterial , and viral proteomes show lack of influence of MHC class I binding preferences ., However , in the same work , A02 and B07 supertypes showed bias in signal-peptide regions ( i . e . N-terminal ) ., We confirmed that using our prediction strategy , signal-peptide specific biases observed for A02 and B07 are present for both human and viral antigens if residue positions are used instead of normalized positions ( data not shown ) ., A DRiP-independent factor that could explain the positional bias of viral epitopes is positional bias of protein conservation ., Specifically , if ends of proteins are less conserved than the middle region and epitopes tend to be more conserved than non-epitopes , positional bias of epitopes may result ., To test this possibility , we calculated conservation scores at the residue-level for proteins of the viruses ( See Methods for details on calculation of conservation scores ) ., Conservation scores could be assigned to ∼60% of proteins ., The remaining proteins had an insufficient number of suitably distant homologues to construct reliable conservation scores , and were therefore excluded from this analysis ., In Fig . 3 , a boxplot showing confidence intervals of means of conservation scores as a function of normalized position is shown ., This reveals a pattern of positional bias of conservation , where ends of proteins are less conserved than their cores , similar to the pattern observed for the positional bias of known epitopes in Fig . 1C ., For the middle bin , however , we do see a difference ., To determine if sequence conservation alone can explain positional bias of epitopes , we first had to determine the relationship between conservation and immune recognition ., In Fig . 4A , distributions of conservation scores for positives and negatives are shown ., Positives tend to be more conserved than negatives ( Welchs t-test; one-sided; p-value\u200a=\u200a5 . 9×10−8 ) ., In Fig . 4B , a plot of probability ratio as a function of conservation score is shown ., The plot is a result of calculating ratios of probability masses for positives and negatives shown in Fig . 4A ., As shown in Fig . 4B , detecting epitopes is less likely as conservation decreases ., Next , we combined our estimates of conservation bias over the length of a protein with our estimate of correlation between conservation and immune recognition ., Using conservation as a function of normalized position in Fig . 3 as input , we estimated probability ratios as a function of normalized position ( Fig . 4C ) ., Its overlay with the probability ratios curve observed earlier ( i . e . Fig . 1C ) is shown in Fig . 4D ., Overall , the probability ratios curve estimated only from the conservation data has a good agreement with the observed one ( Pearsons correlation r\u200a=\u200a0 . 57 ) ., In addition , for first , second and fourth bins , their confidence intervals between observed positional bias curve and that derived from conservation overlap ., Thus , the observed positional bias in viral epitopes can be explained by the correlation between conservation and immune recognition alone , consistent with a minor contribution from premature translational termination and downstream initiation ., There are two possible explanations for the observed correlation between immune recognition and conservation of peptides ., One explanation is that the immune recognition machinery has evolved to preferably recognize epitopes that are conserved , as evidenced by an overlap of MHC binding motifs on conserved sequence regions found in 15 ., An alternative explanation is that viral sequences are variable , and responses against epitopes from conserved regions could be higher in individuals that are exposed to multiple variants of a virus over time , as is expected for example in human influenza infections ., To determine whether there is an intrinsic enhanced immune recognition for peptides that are conserved in viral species , we retrieved viral epitopes identified in the context of peptide immunization , rather than with viral infection ., Figure 5 shows distributions of conservation scores for positive and negative peptides , as was done earlier ., The two distributions overlap , and their difference in means is not statistically significant ( p-value\u200a=\u200a0 . 62 ) ., Thus , we conclude that we cannot detect an intrinsically enhanced immune recognition for peptides that are conserved ., This leads us to favor the hypothesis that the observed enhanced immune recognition of conserved viral peptides is due to extrinsic effects such as repeated priming of responses against conserved peptides due to heterologous exposure ., By leveraging a large set of experimentally determined epitopes from a wide range of viruses stored in the IEDB , we determined positional bias of epitopes in source antigens ., The shape of positional bias curve ( Fig . 1 ) shows significant under-representation of epitopes at N- and C- termini ., This could be explained by near equal participation of downstream initiation and premature termination mechanisms in generating DRiPs as a major source of epitopes ., Our findings , however , point to the conclusion that the central bias of viral epitopes reflects the combined effects of positional bias of epitope sequence conservation and induction of memory CD8+ T cells by exposure to heterologous boosts ., There is a good correlation between positional bias curve estimated only from conservation data with that of the observed position bias curve of epitopes , so we cannot clearly detect ( or clearly rule out ) partially translated DRiPs transcripts as a major source of viral epitopes ., The connection between positional epitope biases with protein conservation is reasonable in the context of boosting effects associated with repeated vaccine administrations ., The principle idea behind boosting is that those epitopes already exposed to the immune system tend to dominate in the following exposure 16 ., This repeated exposure is a likely explanation of why MHC-I restricted viral epitopes tend to be more conserved than non-epitopes ., Providing an additional support to this explanation , Kim et al . presented results showing positive correlation between epitope conservation and T-cell response frequency scores , which indicate how often individuals recognize a given peptide from a pathogen 17 ., Presumably , this correlation is due to higher likelihood of more conserved epitopes being seen by greater number of individuals ., In addition , as more immune epitope data are deposited at the IEDB , we expect to see differences in the degree of positional bias for RNA versus DNA viruses , because RNA viruses are more variable ., In addition to explanations discussed above , there are a number of recent ones that may be relevant ., First , Calis et al . have reported correlations between G+C content and potential MHC-I binders 18 , where low G+C content indicates pathogenicity ., In our hands , we could not detect significant contribution of MHC-I binding affinity preferences to positional bias of epitopes , thereby making G+C content a less likely explanation ., In explaining differential conservations between positive and negative peptides , it may be that this is due to viruses selectively mutating T-cell epitopes 19–21 ., The idea is interesting and may be pursued in a future study ., Our data set however takes protein variability as given and thus cannot be used to delineate its cause/effect relationships ., Other investigators have also reported epitope conservation for HIV 22 , 23 and TB 24 ., Our results extend their findings to a broad set of viruses , and suggest a possible connection between epitope conservation and boosting through analyses of epitopes determined in the context of immunization with peptides ., It remains to be seen whether this boosting effect can be also observed for MHC class II restricted epitopes as well as in other immunological contexts ., Regarding the MHC-motif specific biases and conservation , it has been reported that predicted binding affinities of HLA molecules positively correlate with conserved regions of a wide range of viruses 25 , which appears to contradict the results of absence of MHC-specific biases presented here ., The absence of bias may be explained by the fact that the correlation between predicted binding affinity and protein conservation reported in 25 is very small to begin with ( Spearmans rank correlation of at most ∼0 . 2 ) , thereby dampening any MHC-specific biases that can be seen ., In conclusion , to better understand mechanistic details of antigen processing steps involving DRiPs , positional bias of MHC-I restricted viral T-cell epitopes was measured ., Our findings indicate that there is indeed such bias in antigens , where epitopes at N- and C-termini are under-represented ., Although mechanisms associated with translational errors such as downstream initiation and premature termination may contribute to observed positional bias , our data indicate that differential conservation spanning protein length is an alternative explanation ., For each virus listed in Table 1 , we retrieved MHC-I restricted T-cell epitopes from the Immune Epitope Database ( IEDB ) ( http://www . iedb . org ) , which is the largest publicly available database of epitopes for infectious agents 8 ., To retrieve an appropriate set of epitopes to examine the DRiPs hypothesis , there were a number of important considerations ., First , the query should retrieve epitopes derived from proteins newly expressed in a host cell , rather than epitopes recognized after peptide or protein immunization ., Only for newly synthesized epitopes can defective ribosomal products skew the positional distribution of epitopes in antigens ., To meet this requirement , we query the IEDB for epitopes identified using assays in which the ‘Immunogen Type’ is a whole ‘Organism’ ( rather than an individual peptide or antigen ) ., Second , we further limit the query to epitopes restricted by MHC class I molecules ., Third , we limit the query to epitopes with ‘virus’ as the source organism ., We then grouped the epitopes retrieved with the query by viral species ., As described below , we want to map all epitopes from one species to a single reference proteome ., Therefore , we excluded all viruses for which we do not have reference proteomes available , resulting in a total of 93 different viruses ., To ensure consistent calculation of positional bias , we mapped all epitopes onto antigens from a single complete reference genome for each species based on sequence similarity rather than using the source antigens listed in the IEDB , which are those specified by the author mapping the epitope and are derived from different strains and are of divergent quality ., For example , an author may have used truncated versions of an antigen , or epitopes may come from a polyprotein of Dengue virus , which later gets cleaved into individual products ., Consequently , epitope positions can be made relative to the polyprotein or to final cleavage products ., To carry out the mapping , we used NCBIs BLAST with a default setting to search for presence of epitopes in antigens and to retrieve only those hits with exact matches in the reference genome ., In addition , we required homology between the originally curated source antigen of the epitope and the antigen in the reference genome using BLAST searches with an E-value cutoff of 0 . 001 , thereby ensuring meaningful mapping ., Lastly , we required that there is only a single best match of the epitope in the reference genome to ensure that the position of the epitope in the antigen can be uniquely determined ., We did not consider ties because of associated uncertainty in mapping ., To derive a measure of epitope position that is independent of protein length , a normalized position , x , is defined as follows: ( 1 ) In the equation , ‘peptide_start’ indicates the position of the first residue from the peptide mapped onto the protein sequence; ‘protein_length’ and ‘peptide_length’ are lengths of protein and peptide , respectively ., The first residue of a protein has a position of 1 , rather than, 0 . Our measure of normalized position , x , has the property that if a peptide contains the first residue of the protein ( and thus comes from an N-terminal region ) , then its value is zero; if the peptide contains the last residue , then the value is 1 . 0 ., Consequently , if all regions of a protein are equally likely to contain epitopes , then one would obtain a uniform distribution of normalized positions in the range 0 , 1 ., This property has been verified using randomly sampled peptide positions and uniformly sampled peptides from a set of proteins , followed by mapping the peptides onto the proteins with BLAST ., After mapping peptides with positive ( i . e . epitopes ) and negative T-cell assay outcomes onto their corresponding antigens , we calculated their normalized positions , x , as described in Equation, 1 . We then grouped the normalized positions into ‘positives’ and ‘negatives’ , and binned using 5 bins of equal intervals covering a range 0 , 1 , resulting in probability mass functions ( PMFs ) : p ( x|positive ) and p ( x|negative ) ., The function p ( x|positive ) gives a probability of observing peptides at binned position x , given that only positives were considered ., To indicate positional bias , we calculated ratios of probability masses for positive and negative PMFs: p ( x|positive ) /p ( x|negative ) ., Absence of positional bias corresponds to a probability ratio of 1 . 0 for all bins ., A probability ratio less than 1 indicates under-representation of epitopes while greater than 1 indicates over-representation ., To determine whether differences in conservation over the normalized position in a protein contribute to positional bias of epitopes , we estimated conservations at the residue level for proteins from the viruses using Rate4site algorithm 26 ., We chose the algorithm because it was identified as one of the highest performing methods in a recent benchmark to predict a known set of protein catalytic sites 27 ., To estimate conservation , we used a protocol similar to one used by the ConSurf website 28 ., Specifically , a multiple sequence alignment was generated for each protein by running a sequence of PSI-BLAST→CD-HIT→MUSCLE against the NCBI Non-Redundant database , using a set of Perl scripts retrieved from the ConSurf website ., Running the conservation score estimation algorithm on the alignment returned residue position-specific conservation scores . | Introduction, Results, Discussion, Methods | The immune system rapidly responds to intracellular infections by detecting MHC class I restricted T-cell epitopes presented on infected cells ., It was originally thought that viral peptides are liberated during constitutive protein turnover , but this conflicts with the observation that viral epitopes are detected within minutes of their synthesis even when their source proteins exhibit half-lives of days ., The DRiPs hypothesis proposes that epitopes derive from Defective Ribosomal Products ( DRiPs ) , rather than degradation of mature protein products ., One potential source of DRiPs is premature translation termination ., If this is a major source of DRiPs , this should be reflected in positional bias towards the N-terminus ., By contrast , if downstream initiation is a major source of DRiPs , there should be positional bias towards the C-terminus ., Here , we systematically assessed positional bias of epitopes in viral antigens , exploiting the large set of data available in the Immune Epitope Database and Analysis Resource ., We show a statistically significant degree of positional skewing among epitopes; epitopes from both ends of antigens tend to be under-represented ., Centric-skewing correlates with a bias towards class I binding peptides being over-represented in the middle , in parallel with a higher degree of evolutionary conservation . | To defend the host from an infection , the immune system continuously scans cell surfaces for foreign objects ., Specifically , a virus inside a cell exploits the host to make copies of its proteins; viral proteins are broken up into peptide fragments; and the fragments are displayed on the infected cells surface , thereby allowing detection and cell-killing ., How these peptide fragments for cell-surface presentation are generated remains unknown ., An understanding of this step will lead to rational design of vaccines and insights into tumor immunosurveillance and autoimmunity ., One possible mechanism is that the peptide fragments come from defective proteins missing either the beginning or end regions , which may result in a bias ., Here , we analyzed locations of a large set of known viral epitopes , peptide fragments recognized by the immune system , within their proteins ., We find that all regions of proteins are represented well by the immune system ., However , there is a statistically significant bias in the central regions of proteins , which correlate with a pattern of conservation spanning the length of viral proteins ., Our results suggest a combined effect of conservation and enhancement of immune responses through repeated exposures in shaping the distribution of known viral epitopes . | sequence analysis, biochemistry, proteomics, biology | null |
journal.pgen.1000973 | 2,010 | Sgs1 and Exo1 Redundantly Inhibit Break-Induced Replication and De Novo Telomere Addition at Broken Chromosome Ends | DNA double-strand breaks ( DSBs ) are generated by normal cellular processes including DNA replication or by exposure to DNA damaging agents or ionizing radiation ., To maintain cell viability and preserve genomic integrity , cells employ multiple pathways of homologous recombination ( HR ) to repair DSBs 1–4 ., A key initial step in HR is 5′ to 3′ resection of DSB ends to create single-stranded DNA ( ssDNA ) that recruits formation of a Rad51 filament , which engages in a search for homologous sequences ., The predominant HR pathway is gene conversion ( GC ) , a conservative mechanism in which both ends of the DSB share homologous sequences on a sister chromatid , a homologous chromosome , or at an ectopic location ., Rad51-mediated strand invasion of the 3′-ended ssDNA allows the initiation of new DNA synthesis to copy a short region of the template and patch up the DSB ., When only one DSB end shares homology to a template elsewhere in the genome , a less-efficient HR mechanism , break-induced replication ( BIR ) , can be used to repair the break 5 , 6 ., In BIR , recombination is used to establish an uni-directional replication fork that can copy the template DNA to the end of the chromosome ., If homologous sequences are located ectopically , BIR will result in formation of a non-reciprocal translocation with loss of the distal part of the broken chromosome and may be a significant source of gross chromosomal rearrangements ( GCRs ) and genomic instability 7 ., BIR requires the non-essential subunit of the Polδ polymerase , Pol32 , and all of the essential replication machinery except those excluisvely required for formation of the pre-replicative complex 8 , 9 ., BIR can be used to restart stalled or collapsed replication forks during DNA replication 10 and elongate telomeres in the absence of telomerase 8 ., An alternative way to repair the DSB is through de novo telomere addition through the action of telomerase 11–13 , although this is a very inefficient process that is improved by elimination of the Pif1 helicase 14 ., Genetic and in vivo molecular biological experiments indicate that the early steps of GC and BIR are shared 15–17 ., Following the generation of a DSB , the Tel1/ATM kinase is loaded at sites of DSBs in an Mre11-Rad50-Xrs2 ( MRX ) -dependent manner 18 , 19 ., Tel1 in turn phosphorylates MRX 20 , 21 ., The Sae2 and MRX proteins mediate the initial resection 22 , 23 which is continued via two alternate pathways , one using the Exo1 nuclease and the other employing the multifunctional RecQ family helicase Sgs1 , in concert with Top3 , Rmi1 and the essential helicase/nuclease Dna2 22–24 ., DNA resection is also essential to activate the Mec1-dependent DNA damage checkpoint kinase cascade that triggers a cell cycle arrest , allowing time for the cell to repair the beak prior to mitosis 25 ., Following resection , Rad51-mediated strand invasion of the donor template occurs with similar kinetics , but the initiation of DNA synthesis at the 3′-end of the invading strand is greatly delayed in BIR as compared to GC 16 , 17 ., Recently , Jain et al 16 showed that a “Recombination Execution Checkpoint” ( REC ) delays the initiation of BIR synthesis if a second DSB end has not become engaged nearby on the same template ., It is unclear if the delay in BIR synthesis is due to a restructuring of the strand invasion D-loop and/or the recruitment of BIR associated proteins ., The efficiency of BIR is inhibited by Sgs1 , as there is an increase in BIR in sgs1Δ cells 16 ., Sgs1 also has been shown to disrupt HR intermediates 26 , inhibit homeologous recombination 27–29 , and to dissolve double Holiday Junctions ( dHJ ) to yield noncrossovers 30–32 ., To better understand the role of Sgs1 in BIR , we examined mutations of non-essential genes that either cooperate or act redundantly with Sgs1 in many of its roles in DNA metabolism , including DNA resection ., Here we show that deletion of SGS1 or EXO1 increases the efficiency of BIR whereas overabundance of Sgs1 or Exo1 strongly inhibits it ., Overexpression of Exo1 also inhibits GC ., Deletion of other non-essential factors responsible for DNA resection , TEL1 or SAE2 , modestly increases the efficiency of BIR whereas deletion of MRX impairs BIR ., Additionally , we find that overexpression of Rad51 markedly improves the efficiency of BIR but has little effect on GC ., Finally , we show that Sgs1 and Exo1 redundantly prevent remarkably efficient de novo telomere addition at broken chromosome ends , a pathway dependent on both telomerase and Sae2 ., To study BIR we used the haploid Saccharomyces cerevisiae strain JRL346 ., A galactose-inducible HO endonuclease is expressed to induce a DSB at a modified CAN1 locus approximately 30 kb from the telomere in the non-essential terminal region on Chromosome V ( Ch V ) ( Figure 1A ) ., The HO endonuclease cut site and an adjacent hygromycin-resistant marker , HPH-MX , was integrated into the CAN1 locus , deleting the 3′ portion of the gene but retaining the 5′ portion of the gene ( denoted as CA ) ., A 3′ portion of the gene ( denoted as AN1 ) with 1 , 157 base pairs of shared homology to CA on Ch V was introduced in the same orientation into Ch XI , 30 kb from its telomere ., Prior to HO induction , these cells are canavanine-resistant ( CanR ) because CAN1 is disrupted ., Completion of BIR results in a non-reciprocal translocation that duplicates the donor sequences and the more distal part of the left arm of Ch XI , thus restoring an intact CAN1 gene ., These cells become canavanine-sensitive ( CanS ) and hygromycin sensitive ( HphS ) ., About 20% of cells are viable with 99 . 85% of these cells repairing by BIR and a small fraction by nonhomologous end-joining ( NHEJ ) ., The efficiency of BIR repair allows us to physically monitor the kinetics of repair by PCR , Southern blot and pulse-field gel electrophoresis ( PFGE ) , as described in Materials and Methods ., To compare the effects of mutations on GC , we used the isogenic strain JRL475 ( Figure 1B ) ., The GC strain was modified from the BIR strain by introducing 2 , 404 bp of homology marked by URA3 to the other end of the break ( denoted as 1 , for the 3′-end of CAN1 ) ., The insertion of the URA3-1 sequences also deleted 376 bp in the middle of the CAN1 so there is a gap between the homology shared by the two DSB ends created by HO cleavage ( CA-URA3-1 ) with the donor sequences on Ch XI ( AN1 ) ., Repair by GC results in restoration of the CAN1 gene , rendering cells CanS , but , unlike BIR , the Ch V arm distal to the cut site is retained ., When there is a second end of homology to a DSB break , the cell strongly favors GC over BIR 16 , 17 , 33 , so that after induction of a DSB cell viability increases from 20% in the BIR strain to nearly 70% when there are two ends of homology and GC is used to repair the break ( Figure 1B and Figure 2B ) ., To better understand the role of Sgs1 in BIR , we first measured the viability of sgs1Δ cells after inducing a DSB ( Figure 1A ) ., As previously shown 16 , sgs1Δ cells are 1 . 5 times more efficient in BIR compared to wild type cells ( Figure 2A ) , repairing the break with 33% efficiency ( p<0 . 001 ) ., To confirm that the increase in viability directly correlates with an increase in repair product , we monitored the kinetics of repair using the PCR assay that detects the first 242 bp of new DNA synthesis ., The maximum amount of product detected by PCR ( 18% at 12 hours ) in wild type cells ( Figure 2D ) is comparative to the viability of cells ( 21% ) following induction of the DSB ( Figure 2A ) ., As expected , deletion of SGS1 increased the efficiency of product formation compared to wild type cells ( Figure 2D ) ., Using the previously described BIR system involving the LEU2 sequences 16 we also showed that a helicase-dead allele of Sgs1 34 behaves like the complete deletion of Sgs1 ( Figure S1 ) ., We have previously shown that deletion of sgs1Δ does not increase the efficiency of GC events in which there is perfect homology or when there is a small gap in homology of 1 . 2 kb or less 16 , 35 ., We confirmed that sgs1Δ does not affect the efficiency of GC in the ectopic assay used here ( Figure 1B and Figure 2B ) ., To better understand the role of Sgs1 in BIR , we investigated a number genes that have previously been shown to interact genetically with Sgs1 27 , 29 , 36–41 ., Deletions of MSH6 , MUS81 , YEN1 , RAD27 , ESC2 , DIA2 , YBR094w , or RNH202 did not have a statistically significant effect on BIR when tested for viability after inducing a DSB that can only be repaired by BIR ( Table S1 ) ., However , we found that the other non-essential genes required for 5′ to 3′ resection of DSB ends all affect the efficiency BIR ., A deletion of SAE2 resulted in a slight , but statistically significant , increase in viability ( p\u200a=\u200a0 . 02 ) ., In contrast , deleting subunits of the MRX complex , mre11Δ or rad50 Δ , decreased viability nearly 2 fold ( both p\u200a=\u200a0 . 003 ) ( Figure 2A ) ., The effect of deleting mre11Δ or rad50Δ is consistent with results previously seen in a diploid BIR assay in which a DSB is induced at the MAT locus on Ch III 17 , 42 , but differs from a transformation-based BIR assay that saw no requirement for MRX in BIR 15 ., Because Tel1 plays a role in suppressing gross chromosomal rearrangements and enhances Sae2 and MRX activity in DNA resection 43 we asked if deletion of TEL1 would affect BIR ., Similar to sae2Δ , deletion of TEL1 resulted in a small but statistically significant increase in viability ( p\u200a=\u200a0 . 008 ) ( Figure 2A ) ., Complementation of a tel1Δ strain with the kinase-dead allele 20 partially restored viability to wild type levels ( Figure 2A ) ., The Exo1 nuclease acts redundantly with Sgs1 in DNA resection after the initial trimming of the ends by Sae2 and MRX , although by itself exo1Δ has a minimal impact on 5′ to 3′ resection 22–24 ., Similar to sgs1Δ , deletion of EXO1 ( p\u200a=\u200a0 . 001 ) increased viability nearly 1 . 5 times compared to wild type ( Figure 2A ) ., Also like sgs1Δ , deletion of EXO1 increased the efficency of BIR when measured by PCR ( Figure 2D ) and does not affect the efficiency of GC ( Figure 2B ) ., Plamids overexpressing Sgs1 pYES2-SGS1 44 or Exo1 ( pSL44 ) 45 were expressed under the control of a galactose-inducible promoter on a high copy plasmid ., These overexpression plasmids are denoted as pGAL::SGS1 and pGAL::EXO1 , respectively ., Expression is induced concomitantly with HO induction ., In cells carrying pGAL::SGS1 , the efficiency of BIR decreased 5 fold ( p<0 . 001 ) whereas in pGAL::EXO1 the efficiency of BIR decreased 10 fold ( p<0 . 001 ) ( Figure 2A ) ., Overexpression of these genes did not affect cell viability in cells that lacked an HO cleavage site ( data not shown ) ., Furthermore , we found that Exo1 overexpression inhibited BIR prior to inhibition of new DNA synthesis , by monitoring the kinetics of repair by PCR ( Figure S2 ) ., The strong inhibition of BIR by overexpressing Exo1 depends on the nuclease activity of this protein , as there is no such inhibition when we overexpressed plasmids carrying exo1 mutations that are required for exonuclease activity ( Figure 2C ) ., As shown previously 8 , increasing the homology in our BIR assay more than two fold to 2 , 977 bp increases the efficiency of BIR ( Figure 3C ) ., The increase in homology results in slightly higher viability but does not significantly suppress the effects of overexpressing SGS1 or EXO1 ( Figure 3C ) ., When tested in the GC assay , overexpressing Sgs1 had no effect on viability but overproduction of Exo1 decreased viability by half ( Figure 2B ) ., The initiation of BIR is delayed several hours after the ends of the DSB begin to be resected at a wild type rate of about 4 kb/hr 22 , 46 ., We have also previously shown that the abundance of Rad51 is sufficient to continuously coat only about 10 kb of ssDNA on either side of the break 47; consequently it is possible that excess ssDNA would interfere with forming or maintaining a stable and efficient Rad51 filament that is needed to promote strand invasion and initiation of new DNA synthesis ., Excess ssDNA has been previously shown to interfere with recombination in meiotic cells 48 ., We therefore asked if overexpression of Rad51 would also increase the efficiency of BIR , using well-characterized high-copy plasmids in which RAD51 was expressed under the ADH1 promoter ( pDBL ( RAD51 ) ) 49 or under the PGK promoter ( pSJ5 ) ., Strikingly , overexpressing RAD51 in wild type cells caused a 2 . 5-fold increase in viability ( p<0 . 001 ) when expressed under control of either promoter ( Figure 3D ) ., When we tested the same plasmids in the GC assay we found that there was a slight but not statistically significant decrease in viability ( Figure 3E ) ., These results clearly indicate that Rad51 overexpression preferentially stimulates BIR ., Overexpression of RAD51 in the BIR assay with longer homology further increased the efficiency of BIR ( Figure 3C ) ., We also find that the efficiency of BIR is increased when we tested the kinetics of repair by Southern blot ( Figure 3A ) and PCR ( Figure 3B ) ., However , when normalized to the percent of final product the kinetics of repair are not different from wild type cells ( data not shown ) ., An elevated level of Rad51 increased the viability of sgs1Δ , exo1Δ or tel1Δ cells to the level seen for overexpressed RAD51 alone ( Figure 3D ) , so the effects of RAD51 expression and deleting SGS1 or EXO1 are not additive ., However , overexpressing RAD51 in cells also overexpressing SGS1 or EXO1 did not significantly suppress the inhibition of BIR that is seen with overexpressing SGS1 or EXO1 alone ( Figure 3D ) ., These results could suggest that Sgs1 and Exo1 act prior to the rate-limiting step carried out by Rad51 ., In the case of Sgs1 , it could be in dismantling transient strand invasion encounters; for Exo1 , there is no evident mechanism at this point unless a modest increase in resection 45 would overwhelm excess Rad51 ., We examined a a dramatic 2-fold increase in viability in an sgs1D exo1D double mutant compared to sgs1Δ or exo1Δ alone when tested in the BIR assay ( Figure 4A ) ; however this increase is not in the level of BIR ., Instead , it is due to a dramatic increase in new telomere addition , as described below ., There is in fact no increase in BIR events compared to the single mutants and repair appears to be no better than wild type cells when repair was monitored by PCR ( Figure S3 ) ., As has previoulsy been reported 22–24 , we found that resection is severely impaired in sgs1Δ exo1Δ cells as evident by the persistence of the cut chromosome band seen by Southern blot ( data not shown ) ., Although TEL1 and SAE2 moderately inhibit BIR and are involved in DNA resection like SGS1 and EXO1 50 , deleting TEL1 did not cause new telomere additions at the DSB when ablated in combination with sgs1Δ or exo1Δ nor did deletion of SAE2 in combination with exo1Δ ( Figure 4A ) ., As mentioned above , when we analyzed the viablity of sgs1Δ exo1Δ cells , we found that half of the survivors did not have the CanS HphS phenotype indicative of repair by BIR ( Figure 4A ) ., Instead , the new survivors were HphS but CanR , suggesting that they might have lost the terminal non-essential portion of Ch V distal to the cut site but failed to restore a functional CAN1 locus ., Sgs1 has previously been shown to inhibit homeologous recombination 27 , 29 , specifically the formation of translocations between CAN1 and two highly diverged CAN1 homologs , LYP1 and ALP1 , on Ch XIV 51; these rearrangements might be further elevated by the absence of Exo1 ., Alternatively , given that sgs1Δ exo1Δ severely retards 5′ to 3′ resection , the chromosome end could be stabilized , allowing new telomere addition ., To distinguish between these possibilities , we performed pulse field gel electrophoresis ( PFGE ) on 12 independent CanR HphS colonies , comparing them to the starting strain and a survivor that repaired by BIR ( CanS HphS ) ( Figure 5 ) ., The ethidium bromide-stained agarose gel ( Figure 5A ) shows that the majority of the CanR HphS survivors ( lanes 1–11 ) have a smaller chromosome than the starting ( ST ) strain or one repaired by BIR ( B ) ., ( There is no size difference in Ch V size prior to DSB induction and after BIR because the 30 kb of non-essential region distal to the cut site on Ch V is replaced by a duplication of 30 kb from Ch XI . ), We confirmed by Southern blot that the band remaining at the original position of Ch V is Ch VIII , which is approximately the same size as Ch V in this strain background ( data not shown ) ., One CanR HphS colony ( lane, 12 ) increased in size from the original strain ., These data indicate the CanR colonies are not due to mutations in a restored CAN1 gene , and are therefore not repaired by BIR nor by NHEJ that could have deleted a small region including HPH ., To confirm that none of the CanR HphS colonies were repaired by BIR , we probed with the MCH2 probe that hybridizes proximal to the telomere on Ch XI ( Figure 5B ) ., The MCH2 probe hybridized to sequences on Ch XI in every sample , but only to Ch V in the CanS HphS colony that repaired by BIR ., To determine what sequences of Ch V were retained in the CanR HphS colonies , we next probed the blot with a CAN1 probe that hybridizes to the donor sequences on Ch XI and just proximal ( 1 kb ) to the cut site on Ch V ( Figure 1A , Figure 5C ) ., The CAN1 probe hybridized to sequences on Ch XI in all samples and to Ch V in the starting and BIR strains , but only to three CanR HphS colonies ( 1 , 9 and 12 ) ., This result indicates that at least 1 kb of sequence was deleted in the 9 other CanR HphS survivors ., To determine approximately how much sequence was deleted in the other CanR HphS colonies we probed the Southern blot with a NPR2 probe that specifically hybridizes to Ch V 4 kb proximal to the cut site ( Figure 1A and Figure 5D ) ., In this case , the NPR2 probe hybridized to all CanS samples except lanes 3 , 5 , 6 , and 7 ., When we probed with PRB1 that hybridizes approximately 9 kb proximal to the cut site on Ch V , the probe hybridized to Ch V in all CanS survivors ( Figure 1A and Figure 5E ) ., We also probed the blot with the highly diverged ALP1 and LYP1 sequences on Ch XIV with which CAN1 forms translocations in sgs1Δ cells 51 , but these sequences did not hybridize to the novel chromosome in lane 12 ( data not shown ) ., We have not explored further the structure of this translocation ., Based on our PFGE and Southern blot analysis we conclude that the great majority of the CanR HphS survivors result in a truncation of Ch V after limited resection ., To show if the sequences at the terminus of the truncations are indeed new telomeres , we determined the breakpoint of five independent sgs1Δ exo1Δ CanR HphS repaired colonies by PCR , using a Ch V-specific primer and a telomere-specific primer as previously described 52 , 53 ., As shown in Figure 6 , the presence of a new telomere is indicated by a laddered PCR product ., We then sequenced the PCR product using the Ch V-specific primer ., As shown in Table 1 , all five sgs1Δ exo1Δ CanR HphS colonies have new telomere sequences directly added to the Ch V sequences ., Consistent with the PFGE and Southern blot analysis , the breakpoints were not at a uniform location ., Based on our results , we hypothesize that in the absence of both Sgs1 and Exo1 , a DSB frequently results in a truncated chromosome with newly added telomeres and that these additions can occur at several different sites , often as far as between 1 and 4 kb away from the DSB end ., To confirm that these events are telomerase-dependent , we deleted EST2 , an essential components of telomerase ., As shown in Figure 4A , deletion of EST2 does not affect repair by BIR but eliminates recovery of CanR colonies ., We next asked if NHEJ or HR pathways contributed to de novo telomere formation ( Figure 4A ) ., Telomere addition was not dependent on NEJ1 , which is required for NHEJ ., We next deleted RAD51 , which is required for both BIR and GC ., We confirmed that nearly all BIR is eliminated in sgs1Δ exo1Δ rad51Δ cells but also found a 20% increase in the number of cells with new telomeres ., Although overexpression of RAD51 increased the efficiency of BIR it did not suppress new telomere addition ( Figure 4A ) ., We then tested if the MRX-associated exonuclease Sae2 plays a role in new telomere addition ., Recently , Sae2 and Sgs1 have also been shown to act in parallel telomere processing pathways 54 ., Interestingly , when resection is nearly eliminated by deletion of sae2Δ in combination with sgs1Δ exo1Δ , new telomere addition is eliminated and BIR is significantly reduced ( Figure 4A ) ., When TEL1 was deleted in combination with sgs1Δ exo1Δ there was no change in levels of BIR or de novo telomeres compared to sgs1Δ exo1Δ cells ., It has previously been seen that sgs1Δ exo1Δ cells are defective in GC when tested for the ability to successfully complete MAT switching 23 ., When we tested the viability of sgs1Δ exo1Δ cells in our GC assay there was no discenrable effect on viability ., However , when the phenotypes of the viable colonies were examined only 5% were CanS , which is indicative of repair by GC , while the remaining viabile colonies were CanR , consistent with a truncated chromosome ( Figure 4B ) ., The drastic decrease in GC is consistent with previously published defects seen in sgs1Δ exo1Δ cells ., We analyzed 10 independent CanS colonies by PCR to ascertain if the break was repaired by GC ( Figure 4C ) ., In fact , only 5 of the 10 colonies analyzed ( samples S2 , S3 , S4 , S5 , S8 ) repaired by GC whereas 4 of the colonies repaired the break by BIR ( S1 , S6 , S7 , S10 ) ., One colony ( S9 ) had PCR products consistent with repair by both GC and PCR ., The use of BIR to repair half of the sgs1Δ exo1Δ colonies is consistent with the failure of these cells to activate the DNA damage checkpoint and thus to enter mitosis in the absence of DSB repair ., To verify that that the DNA damage checkpoint was impaired by the lack of normal 5′ to 3′ resection of the DSB ends we microscopically monitored the length of the cell cycle of individual cells plated on YEP-Gal to induce HO endonuclease , from the time that an unbudded G1 cell formed a bud until the dumbbell-shaped mother-daughter pair formed the next bud 55 ., Wild type cells in which the DSB cannot be repaired remain arrested prior to anaphase for approximately 6 cell division times relative to an isogenic strain lacking the HO cleavage site 55 ., In contrast , cells of the BIR strain lacking SGS1 , EXO1 and RAD51 , so that they could not repair the DSB by homologous recombination , show a brief , but significant arrest ., These cells extend the cell cycle 1 . 8 times the length of time of a derivative that lacks the cut site ( 6 . 2 h versus 3 . 5 h ) ., Thus , there is still a brief activation of DSB-induced cell cycle arrest but much shorter than when extensive resection activates Mec1 ., As was the case with CanR sgs1Δ exo1Δ colonies found in the BIR assay , the CanR colonies in the GC assay appear to be chromosome truncations with de novo telomere formation ., PCR analysis showed that the broken chromosomes were truncated at different points proximal of the DSB ( Figure S4 ) ., When representative isolates were tested by PCR as mentioned above we found that consistent with new telomere addition there was a laddered PCR product as seen in sgs1Δ exo1Δ cells in the BIR assay ( Figure S4 ) ., We conclude that eliminating both Sgs1 and Exo1 , by markedly reducing 5′ to 3′ resection and most likely by preventing full activation of the Mec1-dependent DNA damage checkpoint ( see Discussion ) , allows a dramatic increase in new telomere formation , rescuing almost half of all cells suffering a DSB ., In this work we show that the RecQ family helicase , Sgs1 , and the Exo1 exonuclease negatively regulate BIR to maintain genomic integrity ., From the observation that the efficiency of BIR was no greater in sgs1Δ exo1Δ than in a single mutant one might conclude that the helicase/endonuclease ( Sgs1-Rmi1-Top3/Dna2 ) and Exo1 act in the same pathway , but since the sgs1Δ exo1Δ double mutant has such distinctly different phenotypes from sgs1Δ or exo1Δ it is difficult to know precisely why the double mutant does not show an increase in BIR similar to that seen when Rad51 is overexpressed in sgs1Δ or exo1Δ alone ., We note also that other proteins responsible for 5′ to 3′ DNA resection , Sae2 and MRX , do not inhibit BIR in the same fashion; but the behavior of sae2Δ or mre11Δ may be explained by their other important roles in other steps in HR 1 , 3 , 4 ., Sgs1 and Exo1 likely do not act in precisely the same way in inhibiting BIR ., Sgs1-mediated inhibition of BIR may involve unwinding of a nascent strand invasion D-loop , as demonstrated in vitro for the human Sgs1 homolog , BLM 56 , 57 ., In vivo it is clear that the Sgs1 helicase can dismantle strand annealings and strand invasions if the heteroduplex DNA contains mismatches 27–29 ., In meiotic recombination , Sgs1 prevents independent strand invasions of alternative templates 58 , 59 ., If Sgs1 dismantles heteroduplex DNA , we might expect that increased homology between the DSB end and the donor template would lead to a more stable D-loop that would counteract Sgs1 ., Increasing the extent of homology from 1 . 1 kb to ∼3 kb did not significantly change the response of cells to overexpression of Sgs1 ., It is also possible that Sgs1 inhibits the recruitment of some of the BIR-associated proteins ., We note that the effect of deleting Sgs1 or Exo1 is not apparent in a different BIR assay system in a diploid in which nearly all homologous sequences distal to the DSB are deleted 17 , 60; and where there are 100 kb of homologous sequences centromere-proximal to the DSB that can be used to initiate BIR ., However , even in this case , many BIR events fail to retain a marker 3 kb proximal to the DSB , suggesting either that more extensive homology increases BIR or that some more proximal sequences are especially favored in initiating BIR 61 ., Rather than acting on D-loop stability , Exo1 may act on the assembly of the BIR replication fork ., In response to DNA damage or defective checkpoint activation , Exo1 has also been shown to process stalled replication forks and resect nascent strands 62 , 63 ., The mechanism by which Exo1 interferes with fork integrity is unclear; it may be possible that the intermediate steps at which the BIR replication fork is assembled are an Exo1 substrate ., We have previously shown that overexpression of Exo1 increases the rate of resection 45; this has not been tested for Sgs1 overexpression ., A unifying hypothesis would be that BIR is severely limited if resection of the DSB ends is too extensive ., There is a limited amount of Rad51 in the cell ( about 3 , 500 molecules ) , enough to cover continuously about 10 kb of ssDNA 47 ., Although Rad51 will initially form a filament with sequences close to the DSB ( including the relevant “CA” sequences that engage in BIR ) , as resection proceeds the continuous polymerization and depolymerization of Rad51 may leave patches of Rad51 along much of the ssDNA so that by the time BIR is seen , many DSBs will not have a continuous Rad51 filament near the 3′ end to promote the completion of recombination ., Thus , even in wild type cells , overexpressing Rad51 would ensure that there would be a functional filament over the CA sequences and BIR would consequently be more efficient ., Deletions of Sgs1 or Exo1 would partially suppress the problem by slowing down resection ( hence BIR is increased 1 . 5 times wild type ) , although we again note that exo1Δ by itself has little visible effect on resection ., Overexpression of Rad51 is apparently unable to suppress the consequences of overexpressing Exo1 or Sgs1 ., It is important to note that Exo1 overexpression is only effective if nuclease activity is preserved; at least some of Exo1s functions in meiosis are independent of nuclease activity ( N . Hunter , personal communication; L . Symington , personal communication ) ., Increasing homology in our assay does not suppress these effects but further increases in homology may do so , as noted above ., It is possible that overexpressing Rad51 could ensure that the 3′-ended single-stranded DNA was better protected against degradation over the long time required to enact BIR , as previously suggested 64 ., However , we have previously shown that in single-strand annealing where one of the flanking 1-kb homologies is very close to the DSB and the other is exposed only after 6 hr of 5′ to 3′ resection , at least 85% of cells are able to accomplish SSA , which would be impossible if even 1 kb of the 3′-end were degraded in the 6-hr period ., Moreover , SSA was equally possible with and without Rad51 16 , arguing that Rad51 did not provide end-protection to the 3′-ended single-strand ., Eliminating both Sgs1 and Exo1 had a marked defect in completing GC but did not impair BIR so severely ., Because resection is severely impaired in the sgs1Δ exo1Δ double mutant , it is possible that the more severe defect in GC is attributable to the need to resect more than 1 kb of intervening sequence before the “1” end of homology would be single-stranded ( see Figure 1B ) ., However , it is also possible that the difference reflects still another defect in sgs1Δ exo1Δ strains , a failure to activate the DNA damage checkpoint because of a lack of sufficient ssDNA 25 , 65 ., If mitosis is not arrested , then cells that have an unrepaired DSB will proceed through mitosis ., This may lead to the loss of the acentric fragment , as we have shown in other assays 66 , so that only the centromere-proximal DSB end will be inherited ., This situation is not fatal for BIR , which only uses homology on that side of the DSB; indeed previous studies 17 , 67 have shown that BIR may actually increase in a checkpoint-deficient situation whereas GC will be defective ., Thus , even when GC should be possible , half of the HR outcomes of the sgs1Δ exo1Δ GC assay proved to be BIR events ., Strikingly , Sgs1and Exo1 also redundantly inhibit new telomere formation ., In a previous study 12 , when an HO-induced DSB was generated in a rad52Δ strain that could not carry out recombination but had apparently normal 5′ to 3′ resection , only about 1% of cells created new telomeres , and this was only in a situation where a “seed” of T2G4 telomere sequences was located centromere-proximal to the DSB ., In the absence of the T2G4 repeats , new telomeres arose less than 0 . 1% of the time ., The remarkably high level of new telomere formation ( up to 50% of all cells ) must be attributable to the elimination of vigorous resection in the double mutant strain , but it is also likely that the failure to activate the Mec1 DNA damage checkpoint also plays a key role ., Recently , Makovets and Blackburn 68 have shown that the Pif1 helicase , which antagonizes new telomere formation 69 , is phosphorylated in a Mec1-dependent fashion; hence if sgs1Δ exo1Δ block resection and that prevents Mec1 activation , new telomeres should increase ., However , in the assay used by Makovets and Blackburn 68 the level of new telomeres added near an HO endonuclease-induced DSB was only about 2% ., Moreover , Chung et al 60 also find that new telomere addition is much less efficient in cells lacking MEC1 compared to sgs1Δ exo1Δ cells ., Hence , it is likely that the 40–50% level of de novo telomere formation we find reflects both the failure to activate Pif1 when the checkpoint is not strongly activated and the severe block on resection itself ., Apparently de novo telomere formation does not require the recruitment of the MRX-Tel1 complex , as a tel1Δ mutant do | Introduction, Results, Discussion, Materials and Methods | In budding yeast , an HO endonuclease-inducible double-strand break ( DSB ) is efficiently repaired by several homologous recombination ( HR ) pathways ., In contrast to gene conversion ( GC ) , where both ends of the DSB can recombine with the same template , break-induced replication ( BIR ) occurs when only the centromere-proximal end of the DSB can locate homologous sequences ., Whereas GC results in a small patch of new DNA synthesis , BIR leads to a nonreciprocal translocation ., The requirements for completing BIR are significantly different from those of GC , but both processes require 5′ to 3′ resection of DSB ends to create single-stranded DNA that leads to formation of a Rad51 filament required to initiate HR ., Resection proceeds by two pathways dependent on Exo1 or the BLM homolog , Sgs1 ., We report that Exo1 and Sgs1 each inhibit BIR but have little effect on GC , while overexpression of either protein severely inhibits BIR ., In contrast , overexpression of Rad51 markedly increases the efficiency of BIR , again with little effect on GC ., In sgs1Δ exo1Δ strains , where there is little 5′ to 3′ resection , the level of BIR is not different from either single mutant; surprisingly , there is a two-fold increase in cell viability after HO induction whereby 40% of all cells survive by formation of a new telomere within a few kb of the site of DNA cleavage ., De novo telomere addition is rare in wild-type , sgs1Δ , or exo1Δ cells ., In sgs1Δ exo1Δ , repair by GC is severely inhibited , but cell viaiblity remains high because of new telomere formation ., These data suggest that the extensive 5′ to 3′ resection that occurs before the initiation of new DNA synthesis in BIR may prevent efficient maintenance of a Rad51 filament near the DSB end ., The severe constraint on 5′ to 3′ resection , which also abrogates activation of the Mec1-dependent DNA damage checkpoint , permits an unprecedented level of new telomere addition . | A chromosomal double-strand break ( DSB ) poses a severe threat to genome integrity , and budding yeast cells use several homologous recombination mechanisms to repair the break ., In gene conversion ( GC ) , both ends of the DSB share homology to an intact donor locus , and the break is repaired by copying the donor to create a small patch of new DNA synthesis ., In break-induced replication ( BIR ) , only one side of the DSB shares homology to a donor , and repair involves assembly of a recombination-dependent replication fork that copies sequences to the end of the template chromosome , yielding a nonreciprocal translocation ., Both processes require that the DSB ends be resected by 5′ to 3′ exonucleases , involving several proteins or protein complexes , including Exo1 and Sgs1-Rmi1-Top3-Dna2 ., We report that ectopic BIR is inhibited independently by Sgs1 and Exo1 and that overexpression of Rad51 recombinase further improves BIR , while GC is largely unaffected ., Surprisingly , when both Sgs1 and Exo1 are deleted , and resection is severely impaired , half of the cells acquire new telomeres rather than completing BIR or GC ., New telomere addition appears to result from the lack of resection itself and from the fact that , without resection , the Mec1 ( ATR ) DNA damage checkpoint fails to inactivate the Pif1 helicase that discourages new telomere formation . | molecular biology/recombination, molecular biology/dna repair | null |
journal.pcbi.1007201 | 2,019 | The formation of preference in risky choice | Decision-making under risk is ubiquitous in daily activities , such as deciding whether to take an umbrella when the weather forecast predicts 50% chance for rain , or whether to purchase a lottery ticket with a winning probability of 1% ., Such decisions are difficult because the outcomes of the alternatives are only known with some probability , and thus they are subject to risk tradeoffs ., For example , when deciding between a lottery that offers $100 with a probability of 50% and an offer of $40 with certainty , one needs to balance between the appeal of the attractive amount ( $100 ) and the risk of getting nothing ( rather than gaining $40 for certain ) ., Choices between such lotteries were the subject of intensive research in economics and experimental psychology that investigated how humans make risky decisions , starting from the normative Expected-Utility ( EU; 1 ) , followed by random utility models 2 and culminating with Cumulative Prospect Theory ( CPT; 3–6 , see also Transfer of Attention eXchange TAX , for a related type of model 7 ) ., Yet despite the impressive success of CPT in accounting for risky choice data ( e . g . , the dependence of risk-aversion on the magnitude of the outcomes probabilities 8 ) , the theory has been criticized for making assumptions that are inconsistent with capacity limitations of human online information processing , and for not explicating the process by which the preferences are formed 9 , 10 ., Several process theories were developed to account for risky choice ., First , heuristic models , such as Priority Heuristic ( PH ) , suggest that preferences are not formed via a compensatory process of averaging over all outcomes ( like in EU and CPT ) , but rather via a sequential process of comparing the alternatives over one specific attribute ( probability or amount ) at a time , in a specified order , and stopping at the first instance in which a termination criterion is satisfied 9 ., Second , a number of models have relied on the sequential-sampling framework 11–14 , which successfully accounted for choices in perceptual tasks , in order to develop a process model of risky choice ., For example , in Decision Field Theory ( DFT; 15 ) , as attention fluctuates between the alternatives , the preference dynamically evolves by integrating amounts , which are sampled with a frequency that is associated with their ( subjective ) probabilities 16 ., In the Decision by Sampling model ( DbS; 17–19 ) , like in PH , the sampling involves comparisons between the values of the alternatives on a specific attribute ( i . e . , amounts or probabilities , but not both ) ., However , unlike PH , DbS does not assume a fixed order of attribute sampling , nor that the decision is settled at a single comparison , but rather a stochastic sampling , which continues until the accumulated difference of favorable comparisons reaches a decision boundary ., Critically , as opposed to EU or CPT , in DbS the processing takes place within-attributes ( i . e . , comparison between amounts or between probabilities ) ., Finally , in the Parallel Constraint Satisfaction model ( PCS; 20 ) , a compensatory within-alternative process similar to EU ( i . e . , multiplication of amounts and probabilities ) is carried out in a parallel and automatic manner; this process is mediated by a connectionist network of bottom-up and top-down connections ., Although several qualitative predictions of the PCS model have been confirmed 20 , this model has not been tested quantitatively in risky choice ., More recently , a number of studies have relied on eye-fixations during choice between alternatives , to gain insight into the preference formation process ., For example , Krajbich , Rangel and colleagues have shown that an extension of the Drift Diffusion Model ( DDM; 12 , 13 ) , the attentional DDM ( aDDM ) , accounts well for observed preferences between consumer products , food items and 50–50 monetary gambles 21–24 ., To do so , the aDDM assumes that the value of the sampled alternative is modulated by eye-fixations , so that the values of the non-fixated alternatives are attenuated compared with the fixated ones ., In the domain of risky choice , a number of studies have contrasted within-alternative and within-attribute models , and reported partial support for both 20 , 24–28 ., In particular , Glöckner and Herbold 20 analyzed risky choice while monitoring eye-movements , and provided evidence against the PH model and in favor of the PCS and DFT models ( see also 29 for similar results ) ., Finally , in a recent investigation of eye-movements during risky choice , Stewart , Hermens , & Matthews 30 concluded that , while eye-movements contribute to choice preference , this contribution is mostly independent of the values sampled ., In other words , the more one looks at an alternative the more likely s/he is to choose it , independently of the magnitude of amount or probability ., The aim of the current study is to develop and contrast process models of risky choice , which are constrained by the eye movements of participants making decisions ., In particular , we adopt an integration-to-boundary framework , which allows to predict both choices and their decision-time , and we extend the aDDM 21 , 22 , 31 approach to the domain of risky choice ( see also 24 for a recent extension to 50–50 monetary gambles ) ., In this regard , a central question is whether the preferences are formed by integrating global alternative-values , based on multiplicative interactions between amounts and probabilities ( within-alternative processing ) , or by sampling and integrating attribute-comparisons ( within–attribute processing ) ., Furthermore , using process models that include attentional modulation of fixated information , we wish to account for individual differences in risk preference ., While previous work has highlighted the impact of task-complexity ( e . g . , number of alternatives and attributes ) in determining the decision strategy adopted by the participants ( e . g . , 32 ) , here we focus on the simplest type of risky choice ( between pairs of alternatives , each consisting of a probability p to win amount x , see Fig 1A ) ., Thus , our aim here is not to determine which of these two types of processes prevail in any choice scenario ( both can take place , subject to task-conditions and individual differences ) ., Rather , we wish to test if , at least for this simple case , the more normative ( within-alternative and multiplicative ) strategies are within the capacity of participants resources ., Towards this end , we carry out a systematic investigation of risky choice with simple two-outcome lotteries , while eye-fixations are monitored ., To anticipate our results , we provide a clear demonstration that within-alternative and multiplicative evaluations are being used , subject to individual differences that correlate with choice normativity ., We began by examining the basic psychometric properties of our choice-data ., Analysis of the catch-trials showed that the participants chose the better option ( higher in both amount and probability ) in 97% of these trials ( SD = 6% ) ., Next , we conducted a mixed-effect logistic regression on the choice data , with the Expected-Value ( EV ) differences ( x1∙p1 –x2∙p2 ) as a predictor , and with random intercepts and slopes at the participant level ., The results indicated that the participants were sensitive to EV differences , and preferred lotteries with higher EVs over lotteries with lower ones ( β = 0 . 40 , p < . 001; Fig 2A ) ., Additionally , using a Pearson correlation analysis , we showed that the reaction time ( RT ) of a decision decreased as the absolute EV difference between the lotteries increased ( r = -0 . 8 , p < . 001; Fig 2B ) ., This finding is consistent with previous process models such as the PCS 20 , the aDDM 21 , and the DFT 16 , indicating that the participants take longer to decide when the evidence ( as measured by the EV-difference ) is smaller ., Finally , we evaluated the risk-preferences of the participants ., To this end , we focused on choice problems with similar EVs ( |ΔEV| ≤ 1 , Nchoice problems = 26 ) , and examined the proportion of trials in which high-payoff/low-probability lotteries ( riskier options ) were preferred over low-payoff/high-probability lotteries ( safer options ) ., Following the CPT regularity of differential risk-attitudes for low vs . medium/high probabilities ( see S1 Text ) , we examined the risk-preferences separately for these two probability domains:, i ) low-probability cases , in which one of the lotteries has p < . 25 ( e . g . , $24 with p = . 1 vs . $6 with p = . 5 ) , and, ii ) high-probability cases , in which both lotteries have p ≥ . 25 ( e . g . , $30 with p = . 5 vs . $15 with p = 1 ) ; the . 25 cutoff was selected to match CPT ( see S1 Text ) ., A paired samples t-test indicated that , consistent with CPT , the participants showed higher levels of risk-aversion for medium/high probabilities as compared to low ones ( t ( 30 ) = 3 . 84 , p < . 001 ) ., Follow-up one-sample t-tests ( against . 5 ) indicated that the participants showed risk-aversion for medium/high probabilities ( t ( 30 ) = 4 . 49 , p < . 001 ) ; no risk-aversion , however , was obtained for low probabilities ( t ( 30 ) = -0 . 11 , p = . 9 ) ., On average , the participants made 9 . 05 fixations ( SD = 3 . 56 ) per trial , with a mean duration of 407ms ( SD = 244 ms ) per fixation ., Also , on average across participants , there was no significant difference between the proportion of fixations towards amounts and probabilities ( t ( 30 ) = 0 . 78 , p = . 44 ) ., There was , however , a remarkable difference between participants in this proportion , which was correlated with participants’ risk preferences: the more a participant fixated on amounts , the more likely he or she was to choose the riskier alternatives ( r = . 48 , p = . 006; S1A Fig ) ., To understand this relationship we examined individual differences in fixating the higher of two amounts/probabilities , as this can explain risk-biases ( looking more at higher amounts or at lower probabilities leads to risk-seeking according to the aDDM 21 , 22 , 24 ) ., Importantly , we find that the more a participant tends to fixate on amounts the more s/he fixates on the larger of them ( r = . 47; p = . 007; S1B Fig ) , and similarly for probabilities ( r = . 46; p = . 007; S1C Fig ) ., Finally , the frequency of fixations on the higher of the two amounts was positively correlated with risk-seeking ( r = . 58; p < . 001; Fig 2E ) , and the frequency of fixations on the higher of the two probabilities was negatively correlated with risk-seeking ( r = . 45; p = . 01;S1D Fig ) see also 24 , 33 ., We also examined the eye-trajectories in relation to their transitions between the four attributes ( x1 , p1 , x2 , p2 ) ., The transitions between decision attributes ( amounts and probabilities ) were classified into three categories 20 , 25 , 30:, i ) Within-alternative transitions–transitions between attributes that belong to the same alternative ., ii ) Within-attribute transitions–transitions between different alternatives , within the same attribute ., iii ) “Diagonal” transitions–transitions between the amount of alternative A and the probability of alternative B and vice versa ., Fig 2C and 2D show one example each for within-alternative and within-attribute trials , respectively ., An Analysis of Variances ( ANOVA ) revealed significant differences of the transition probabilities between the three transitions types ( F ( 2 , 60 ) = 431 . 1 , p < . 001 ) ., Post-hoc comparisons showed that the participants made more within-alternative than within-attribute transitions ( p < . 001 ) , as well as more within-attribute than diagonal ones ( p < . 001 ) ., The proportion of within-alternative transitions ( out of all transitions ) was subject to individual differences and was correlated with EV-choice performance ( ΔEV ) , such that the higher the fraction of within-alternative transitions the higher was the proportion of the alternative with the higher EV to be chosen ( r = . 57 , p < . 001; Fig 2F ) ., Recent research has demonstrated that attentional mechanisms play a key role in the development of preferences 24 , 34–38 ., In particular , it was shown that the more an alternative is fixated on , the more likely it is to be chosen 21 , 30 , 39 ., We first estimated the benefit of looking time ( or of the number of fixations ) on choice , using a measure that was developed in another study ( 40 ) ., Specifically , using logistic regression we predicted the choices from the EU-difference between the two alternatives ( with parameters fitted for each subject on all his/her choices ) ., Then , for each trial , we computed the deviation between the actual choices ( coded as 0 or 1 ) , and the probabilities predicted by the EU-difference ., This residual was averaged separately for trials in which the choice had a positive or a negative gaze-advantage ( we did this twice , for total gaze duration and for number of fixations ) ., We then computed the difference between these measures to obtain the average difference in choice probability for the items with a positive versus negative final gaze advantage , when corrected for the influence of their ( EU ) values ., As shown in Fig 3 ( see also 40 for similar results on non-risky choices ) , there is a marked gaze-advantage in predicting the choice ., At the group level , this advantage has a mean value of 0 . 29 ( SD = 0 . 14 ) for the number of fixation ( Fig 3 , upper panel ) , and a mean value of 0 . 24 ( SD = 0 . 11 ) for the total gaze duration ( Fig 3 , middle panel ) ., Finally , we found a small , but significant prediction enhancement , for the number of fixation predictor ( t ( 30 ) = 2 . 87 , p = 0 . 008; Fig 3 , lower panel ) ., Second , we have confirmed the impact of gaze on choice , using a multiplicative computation of the alternatives values based on EU ( or CPT ) values and a monotonic function of gaze-time ( or number of fixations; see S2 Text for details ) ., As illustrated in Fig 4A , we examined an EU × time regression model ( similar conclusions were obtained for CPT based models , see S2 Text ) , in which the EU value of each alternative increases with its dwell time on the two alternatives ( α is the risk-parameter of EU , τ is a saturation parameter , and β is a slope parameter ) ., Additionally , we examined a similar regression model , in which dwell-times were replaced with the number of fixations each alternative is sampled ( Fig 4B ) ., Comparison of these models with the traditional EU ( which does not take eye-movements into account ) showed that using eye-movements significantly improved prediction accuracy and AIC comparted with the traditional EU ( Fig 4C ) ., Note also , that the prediction accuracy and AIC which were obtained using the number of fixations , equal ( for EU ) or surpass ( for CPT ) , the prediction accuracy and AIC obtained using the measure of dwell-time ., In addition , in both gaze based regressions ( number of fixations and dwell time ) , the fitted values of saturation-parameter τ were lower than 1 , indicating that , for example , looking twice as long at an alternative increases its value by a factor of less than 2 ., One way to understand this non-linear saturation is in relation to a leak of the accumulated values 14 , 41 , 42 ., In such leaky integration models , the accumulated evidence saturates at an asymptotic value , and remains constant even if more integration time is allowed ., Accordingly , at each fixation one samples and accumulates a value , however , as the trial proceeds the accumulated value leaks , resulting in a type of recency ., Indeed , the percentage of match between the fixated alternative and the final choice as a function of fixation number ( backwards from the end ) showed a clear recency pattern ( Fig 5; note that an aDDM model without leak can also generate a recency pattern 43 , therefore a quantitative model comparison is needed to determine if leak is required to account for the actual pattern ) ., The central aim of this study is to develop and contrast two classes of process models that differ in the way attentional ( or eye ) transitions affect the integration of amounts and probabilities ., Both types of models assume that:, a ) fixated objects receive enhanced attention ,, b ) attention modulates the weight of value integration 21 , and, c ) recently sampled values are weighted more than earlier ones 14 , 41 , 42 ., The models differ , however , on how the values are integrated into preferences ., Note that we do not aim to test specific models but rather distinguish between broad classes of models based on certain principles , in particular , between within-attribute vs . within-alternative models 20 , 25 , 32 , 44 ., While the former is used in models such as PH and DBS , the latter is used in models such as EU , CPT and PCS ., We also examined a more hybrid model , which still relies on multiplicative within-alternative computations , but also allows some extent of competition between the attributes ., The most complex of the models ( in terms of number of parameters ) is the within-alternative process model , which has four free parameters ., The first two , α and γ correspond to the CPT parameters for risk aversion and probability weighting 3 , respectively , θ corresponds to the aDDM attentional modulation parameter , and λ is the activation-leak ., As we show in S5 Fig , we carried out a recovery exercise , showing that our fitting procedure is able to provide a good recovery for all those parameters over a wide range of values that correspond to those found in the actual data ., This non-trivial result is helped by the fact that our 94 choice problems systematically span the choice space ., As shown in Table 1 , the within-alternative process models with attention modulation and leak gave the best fit and showed the highest cross-validation prediction accuracy ., They outperformed both the within-attribute process models , as well as the traditional , non-integration to boundary models ( compensatory and non-compensatory heuristics ) ., These results speak against the hypothesis that the participants accumulate only the differences of the attended attributes ., We also found that the within-attribute models with perfect ( rather than leaky ) integration ( Normalized and Binary differences ) , resulted in much worse AIC , prediction accuracy , and cross-validation ( therefore in Table 1 we report only the within-attribute models which include leak as a free parameter ) ., We note that the within-alternative choice models required a significant degree of information leak ( λgroup = 0 . 58 ) ., As shown in S3 Table , we explicitly tested four versions of within-alternative models that include an attentional modulation but no activation-leak , all of which resulted in much poorer prediction-accuracy and AIC fit values ., By contrast , the leaky within-alternative process models outperformed ( on prediction accuracy , AIC and cross-validation ) the regression models that include either EU or CPT together with the number of fixations ( see S2 Text and Table 2 ) ., This suggests that considering dynamic processes , such as attentional shifts and leak of activation improves prediction accuracy and fit measures beyond what is achievable by using only the number of fixations ., Note that the within-alternative two-layer leaky accumulators model outperforms the single-layer accumulator model ., This result suggests that the perception of the attributes is dynamic and is subject to modulation by attentional processes ., Finally , the hybrid model resulted in fits ( AIC and prediction accuracy ) that did not exceed those of the within-alternative model ( see S3 Table ) , and with a moderate mutual inhibition value ( . 13 ) which does not trigger a full all-or-none dynamics ., Due to its complexity , we leave a full investigation of this model to future research ., Finally , we carried out a comparison of the predictive accuracy of our best performance model–the two-layer leaky accumulators—with that of the traditional EU and CPT models across all decisions as a function of EV-differences ., The comparison demonstrates that the difference in prediction accuracy is especially large for difficult choices ( low EV-differences , 1–3 Quantiles; Fig 7 ) , suggesting that attentional modulations are particularly significant in difficult decisions 48 ., We contrasted the within-alternative and the within-attribute models , in accounting simultaneously for choices and decision-times ., To this end , we adopt an integration-to-boundary framework , which assumes that preferences are accumulated until they cross a decision criterion 49 , 50; this introduced a few more parameters ( for the boundary ) into the model ( see S1 Methods Model Fitting ) ., The models are now set to estimate the probability of a subject’s choice conditioned on its decision time and fixations ., This probability is accumulated for all choice trials of the participant to a total likelihood , which is used to optimize the boundary parameters ., Two families of decision boundaries were tested , for each of the models:, i ) the standard fixed ( time-invariant ) boundary , which introduces a single new boundary parameter , and, ii ) a collapsing ( time-variant ) boundary model , which introduces three new parameters ( see S1 Methods Optimization procedure: choices and decision-times , for further details regarding the implementations of these two types of models ) ., The collapsing boundary model has been the focus of recent investigations in decision neuroscience 41 , 42 , and appears to be favored in experimental tasks that span over longer time intervals ( more than 2–3 sec 51–53 , but see 54 for an alternative explanation based on across-trial variability parameters ) ., The results show that , with both decision boundary families , the two-layer leaky accumulator model outperformed all the other models ., Among the two types of boundary families , the best fits were obtained for the collapsing boundary models ( AIC and cross-validation ) , despite the cost of the two extra parameters ., For this reason , we only report below the results for this type of boundary ( see S4 Table for comparison of all within-attribute and within-alternative models using fixed and collapsing boundaries ) ., We find that the within-alternative/two-layer leaky accumulator model ( AIC = 14 , 492 ) decisively outperformed the within-attribute/normalized differences model ( AIC = 15 , 815; ΔAIC = 823 ) , in accounting for decision-times ( conditioned on the actual fixation patterns ) ., Finally , we used these models to predict the distribution of decision times ( measured in number of fixations ) , for novel but statistically matched patterns of fixations ., To this end , for each trial we simulated a fixation sequence that is based on a statistical model of the participant’s fixations towards the four attributes as a function of their values 21 , 30 ., The results indicate that for the two-layer leaky accumulator model , the predicted and actual decision-time distributions show a good match , however for the normalized differences model , the tail of the predicted decision-time distribution deviates from that of the actual decision-time distribution ( Fig 8 ) ., Our best within-alternative integration model accounts also for the empirical correlation we reported between the proportion of fixations a participant makes to the higher of the two amounts and his or her risk-preference bias ( Fig 2E; see also 24 ) ., To show this , we simulated choices for each participant , based on his or her fitted model-parameters and the participants actual fixation sequence ., The correlation between the models risk-preference prediction and the proportion of fixations to the higher amount ( r = . 58 , p < . 001 ) , was exactly equal to the empirical correlation obtained in the data ( Fig 2E ) ., Next , we sought to demonstrate that this relation is associated with the fixation pattern and not merely with differences in model parameters ., To this end , we simulated choices for each participant , by using his or her actual fixation sequences , however , this time we used model parameters that correspond to the group mean ( rather than the individually fitted parameters ) ., This resulted in a significant correlation ( r = . 52; p = . 002; Fig 9A ) between the risk-preference and the proportion of fixating on the higher amount ., This correlation between risk-biases and fixation-pattern relies upon the model’s attentional component , which gives higher weights to the attributes on which the participant fixates ., For example , assume that a participant is asked to choose between A: ( $20 , 0 . 5 ) and B: ( $10 , 1 ) ., If s/he fixates more the amount of alternative A than the amount of alternative B , higher weights would be given to the former , and thus the riskier alternative ( A ) would be preferred by the model over the safer one ( B ) ., Finally , we address an important question: which preference-formation mechanism ( within-alternative or within-attribute ) results in better normative performance , and thus can be regarded as more adaptive ?, To answer this , we simulated the two types of models based on the participants best fitted parameters and actual fixation sequences , and we examined two measures of normative choice predicted by each model:, i ) the fraction of EV-choices ( for simplification we discuss normativity in terms of EV , but the same would hold in terms of EU ) , and, ii ) the fraction of transitivity violations–a direct measure of choice irrationality ( 55 , 56; see S4 Text ) ., As seen in Fig 9B , the normative performance is higher for the within-alternative model than for the within-attribute model , for both measures: EV-choices: t ( 30 ) = 6 . 27 , p < . 001 and transitivity violations: t ( 30 ) = 5 . 15 , p < . 001 ., This is expected because our within-alternative model , like CPT , assumes a multiplication between subjectively transformed amounts and probabilities , which also maintains choice-consistency ., Although the normative model requires a multiplication of objective values whereas our model requires a multiplication of subjective values , this discrepancy is relatively minor compared with non-multiplicative strategies ( i . e . , within-attribute integration or heuristics ) ., Moreover , we have found that the more within-alternative transitions a person makes , the higher is his or her fraction of EV choices ( Fig 2F; see also 28 ) ., This correlation can be naturally understood , since the participants rely on a within-alternative multiplicative mechanism , and this operation is likely to be more precise following an actual transition between amounts and probabilities ( i . e . , a fixation on one attribute of an alternative followed immediately by a fixation to the other attribute of the same alternative ) , than following a non-direct transition ( where one of the to-be-multiplied attributes is based on memory or defaults ) ., Consistent with this , we found a correlation across participants between the prediction accuracy of the within-alternative model and the proportion of within-alternative transitions ( r = 0 . 39 , p = . 03 ) ., The main aim of our study was to elucidate the mechanisms by which different attributes ( amounts and probabilities ) are integrated to generate an overall subjective value of choice alternative ., To this end , we focused on choices between simple lotteries and developed process models of risky choice , which are constrained by eye-fixations and we assumed a fixation-based attentional modulation ., In addition , we introduced activation-leak and examined two types of decision-boundaries , in order to account for decision times ., Within these models we specifically contrasted within-alternative multiplicative models and within-attribute type models , and carried out a systematic parametric investigation of choices between simple lotteries ( x1 with p1 vs . x2 with p2 ) , while tracking participants eye-fixations ., First , we replicated previous findings indicating that participants prefer lotteries with higher EVs ., In particular , the choice probability of the alternative with the higher EV increases ( and choice-RT decreases ) with the EV-difference between the lotteries ., Nevertheless , participants also exhibited risk biases that are probability-dependent , being risk-averse at high/medium probabilities , but not at low probabilities ., Second , we found that , on average , the eye-scan patterns were dominated by within-alternative as compared to within-attribute or diagonal transitions ( Fig 2C–2D , respectively ) , and that individual differences on this eye-scan pattern correlate with EV-choice ( see also 28 , for a similar result ) ., Third , we used eye-fixations to constrain a number of process models that accumulate preference across fixations , using an aDDM approach with two attributes 21 , 57 ., Here we contrasted two types of integration-to-boundary process models:, i ) within-attribute models , and, ii ) within-alternative models ., As shown in Table 1 , the latter resulted in the best predictive accuracy and measures of fit ., Importantly , the two-layer model also accounted well for decision times ( Fig 8 ) and for individual differences in risk biases ( Fig 9 ) ., Finally , the worst performance in our task was obtained for the non-compensatory heuristic models ., For example , the best of the heuristics ( the PH ) resulted in a worse fit than even the simple EV model ( see cross-validation measure in Table 1 ) ., The conclusions favoring the within-alternative multiplicative models may need to be qualified for the task conditions we used here ., First , we used simple lotteries with single non-zero outcomes ( x with p , 0 with 1-p ) ., It is possible that the amount of within-attribute ( non-compensatory ) processing would increase when more complex choices are used 25 , 32 ., While we cannot rule out this possibility , recent research in the domain of probabilistic inferences 58–60 and risky choice 20 , 61 , indicates that when decision processes are monitored via eye-tracking ( which does not slow down the decision process ) rather than via mouse pressing techniques ( e . g . , 10 ) , participants are able to use compensatory strategies for relatively high complexity levels ( see also 62 for a multi-attribute choice task ) ., Second , our results need to be qualified to the use of analog ( graphic rather than symbolic ) representation of the data ., We used this format of representation to reduce the possibility that our participants ( who are students that may be familiar with EV-principles , and are required to do 104 choice problems ) , adopt an explicit EV calculation strategy ., We believe that such a strategy is less likely with analog information , and thus our results favoring an implicit multiplicative mechanism are even more remarkable ., Third , the alternatives in our experiment were aligned only vertically ( one lottery was placed over the other , with the amounts and probabilities placed left/right , see Fig 1A ) ., It is possible to argue that this layout favors within-alternative processing ( left to right or right to left transitions ) ., Note , however , that a horizontal layout ( left/right alternatives ) triggers a strong bias favoring within-attribute processing , in particular , comparing the horizontally aligned amount bars ., Nevertheless , we report in S4 Fig data from a pilot Experiment ( N = 13 ) using a horizontal alternative layout , which shows that even under such within-attribute favorable conditions , we still find dominance for within-alternative transitions ., Here we only wish to support the following conclusion: humans possess the capacity to spontaneously deploy an ‘economic’ ( multiplicative across-dimension ) type computation ( which is analog rathe | Introduction, Results, Discussion, Methods | A key question in decision-making is how people integrate amounts and probabilities to form preferences between risky alternatives ., Here we rely on the general principle of integration-to-boundary to develop several biologically plausible process models of risky-choice , which account for both choices and response-times ., These models allowed us to contrast two influential competing theories:, i ) within-alternative evaluations , based on multiplicative interaction between amounts and probabilities ,, ii ) within-attribute comparisons across alternatives ., To constrain the preference formation process , we monitored eye-fixations during decisions between pairs of simple lotteries , designed to systematically span the decision-space ., The behavioral results indicate that the participants eye-scanning patterns were associated with risk-preferences and expected-value maximization ., Crucially , model comparisons showed that within-alternative process models decisively outperformed within-attribute ones , in accounting for choices and response-times ., These findings elucidate the psychological processes underlying preference formation when making risky-choices , and suggest that compensatory , within-alternative integration is an adaptive mechanism employed in human decision-making . | Decision-making under risk requires a selection between alternatives , such as lotteries , which offer a reward with a specified probability ., Human decision between such alternatives is at the center of the normative decision theory , which assumes that decisions are rationally made by forming a value for each alternative and selecting the alternative with the highest value ., To this day , there is still a considerable debate on how such values are computed ., While the normative theory assumes that values of the alternatives reflect the statistically expected rewards , more recent theories have argued that alternative-values are not computed , and choices are only based on sequentially comparing the alternatives on amounts or on probabilities ., Here , we carried out an experimental investigation of risky decision-making , in which participants chose between pairs of simple lottery alternatives that systematically span a range of probabilities and amounts , while we tracked their eye positions during the decision-making process ., We found that the participants are sensitive to the expected-utility of the alternatives , as predicted by the normative decision theories , but they also exhibit risk-biases that correlate with the eye-scanning patterns ., We then carry out computational modeling , comparing preference-formation models on their ability to account for both choices and their reaction-times ., The results provide strong support for normative models , which assume that the values of the alternative are computed via a multiplicative function of the amounts and probabilities ., These results suggest that humans are closer to normative principles than previously assumed , and motivate further investigation into the neural mechanism that mediates these multiplicative computations . | infographics, medicine and health sciences, decision theory, decision making, statistics, applied mathematics, social sciences, neuroscience, research design, cognitive psychology, mathematics, cognition, research and analysis methods, sensory physiology, computer and information sciences, experimental economics, economics, pilot studies, probability theory, visual system, psychology, eye movements, data visualization, physiology, graphs, biology and life sciences, sensory systems, physical sciences, cognitive science, attention | null |
journal.pgen.1008045 | 2,019 | Discovery and characterization of variance QTLs in human induced pluripotent stem cells | Robustness , or the ability to maintain a stable phenotype despite genetic mutations and environmental perturbations , is an important property of many key biological processes , such as those underlying embryogenesis and development 1 , 2 ., Conversely , evolvability , or the ability to generate heritable phenotypic variation , is a fundamental requirement of evolutionary processes 3 ., A long-standing question in genetics , therefore , is how the balance between these two seemingly opposite processes has been fine-tuned 4 ., To make progress in understanding the balance between robustness and evolvability , we need to characterize the mechanisms that underlie robustness ., Robustness can arise through a number of different mechanisms: for example , redundancy of regulatory elements or feedback loops in regulatory circuits ., In these different scenarios , we hypothesize evolvability could be maintained through different selective pressures ., If we are able to characterize gene-specific robustness to expression variability , we can begin to ask about the balance between natural selection of gene function and the ability to maintain evolvability ., In model organisms , robustness and evolvability can be studied using experimental evolution approaches ., These approaches typically quantify robustness as the change in trait variation after applying an experimental perturbation 5 , 6 ., However , in such experiments the phenotypic outcomes , rather than the underlying mechanisms of robustness , are measured ., Moreover , experimental evolution studies have almost always considered population-average measurements of phenotypes using entire organisms , tissues , or cell cultures , with few exceptions 7 , 8 ., To truly understand how robustness and evolvability are established and encoded in the genome , we need to consider phenotypic variation across individual cells 9 , and connect it to genetic variation , an approach termed “noise genetics” 10 ., Using the yeast Saccharomyces cerevisiae as a model system , studies have shown that heterogeneity in the expression of certain genes across cells is highly heritable and placed under complex genetic control , suggesting that the level of noise in gene regulation may also differ between individuals of multicellular organisms depending on their genetic background 11 ., Follow-up studies further demonstrated that gene expression noise mediated by promoter variants can provide a fitness benefit at times of environmental stress in yeast , highlighting the direct role of genetically controlled stochastic cell-cell variation in evolutionary robustness 12 ., However , the genetic and molecular circuits that lead to robustness remain largely uncharacterized in mammals ., Here , we take an unbiased , genome-wide approach to identify quantitative trait loci associated with gene expression variance across cells ( vQTLs ) ., We study human induced pluripotent stem cells ( iPSCs ) , which offer a homogeneous population of cells allowing a relatively simple statistical model ., Investigating iPSCs also provides the possibility to study gene expression variance across cells during differentiation in follow-up studies ., To directly measure the mean and variance of gene expression within cell populations as phenotypes , we generated single cell RNA-seq ( scRNA-seq ) data from cells derived from multiple individuals ., Using the Fluidigm C1 platform , we isolated and collected scRNA-seq from 7 , 585 single cells from iPSC lines of 54 Yoruba in Ibadan , Nigeria ( YRI ) individuals ., We used unique molecular identifiers ( UMIs ) to tag RNA molecules and account for amplification bias in the single cell data 13 ., To estimate technical confounding effects without requiring separate technical replicates , we used a mixed-individual plate study design ( Fig 1A ) ., The key idea of this approach is that having observations from the same individual under different confounding effects and observations from different individuals under the same confounding effect allows us to distinguish the two sources of variation 14 ., We excluded data from one individual ( NA18498 ) with evidence of contamination , then filtered poor quality samples as previously described 14 ., After quality control , we analyzed the expression of 9 , 957 protein-coding genes in a median of 95 cells per individual in 53 individuals ( total of 5 , 597 cells; S1 Fig ) ., To ensure that our measurements are comparable across samples , we first sought to assess the impact of observed technical variation on the data and to identify unobserved technical confounders ., To this end , we performed principal components analysis ( PCA ) on the matrix of log counts per million ( log CPM ) ., We found that across samples , the loading on the top principal component ( PC ) was correlated with gene detection rate ( the proportion of genes with at least one molecule detected ) , but not with the biological variable of interest ( individual ) or the expected technical confounders ( batch and C1 chip; Fig 1B ) ., Indeed , as previously reported 15 , the entire distribution of observed log CPM ( over all genes ) varies across samples , and appears to be associated with the gene detection rate ( Fig 1C ) ., After accounting for gene detection rate ( Methods ) , the top PCs were correlated with individual , batch , and C1 chip , as expected ( Fig 1D ) ., We developed a method to estimate the mean and variance of gene expression across cells for each gene in each individual ( Fig 1A; Methods , S1 Text ) ., Briefly , for each individual and each gene , our method uses maximum likelihood to fit a zero-inflated negative binomial distribution ( ZINB ) to the observed UMI counts across cells , and derives the mean and variance of gene expression from the estimated model parameters ., When fitting the ZINB model the method controls for technical confounders ( e . g . C1 chip ) and library size , and when deriving the mean and variance it accounts for Poisson measurement noise in the UMI counts 16 , 17 ., These desirable , and arguably crucial features would not be achieved by directly computing the sample mean and variance of either the UMI counts or log CPM ., To evaluate the accuracy of the method , we first simulated data from the model and compared the estimated parameters , as well as the derived mean and variance , to the true values used to generate the data ., We fixed the number of cells and number of molecules detected per cell to the median of those values in our observed data , and varied the ZINB parameters ., Assuming that mean expression is high enough , we found the method produces accurate estimates of the underlying negative binomial parameters , but not the zero inflation parameter ( S2 Fig ) ., Despite not accurately estimating the zero inflation parameter , the method still produces accurate estimates of the derived mean and variance for genes that are expressed at intermediate to high levels ., Next , we tested for goodness of fit on each simulated data set ( Methods ) ., The key idea underlying the test is that if the data are truly distributed according to some cumulative distribution function F , then the values of F evaluated at the data should be uniformly distributed between 0 and 1 ., Applying the test to the simulated data , we rejected the null that the model fit the data for zero of 2 , 451 simulation trials after Bonferroni correction ( p < 2 × 10−5; S3 Fig ) ., The results suggest the method is successfully able to fit the observed data , and also suggest that inaccuracy in the estimated parameters is likely explained by noise due to small sample sizes ., We then applied our method to the observed data , correcting for batch and C1 chip ., Importantly , we did not correct for gene detection rate , reasoning that the dependence on gene detection rate is only an artifact introduced by analyzing log CPM ., We tested the goodness of fit for each individual and each gene , and rejected the null that the model fit the data for only 60 of 537 , 658 individual-gene combinations ( 0 . 01% ) after Bonferroni correction ( p < 9 × 10−8; S4 Fig ) ., Our results emphasize that careful experimental design as well as careful statistical modeling are required to robustly map effects on gene expression variance across cells ., Previous studies have shown a clear relationship between the mean and variance of gene expression 18 , 19; therefore , apparent genetic effects on the variance could potentially be explained by effects on the mean ., In our model , the mean-variance relationship is controlled by a single dispersion parameter per gene per individual ., We sought to directly map QTLs which could alter the variance independently of altering the mean by using the estimated dispersion parameter as a quantitative phenotype ., However , we found zero dispersion QTLs ( dQTLs ) using this approach ( FDR 10% ) ., Further , we found the QQ plot of association p-values did not show deviation from the null ( S5 Fig ) ., Alternative approaches to decouple the mean-variance relationship include using the coefficient of variance ( CV; ratio of standard deviation to mean ) or Fano factor ( ratio of variance to mean ) as quantitative phenotypes ., However , prior work shows these quantities have predictable relationships with the mean , and therefore effects could still be explained away 14 , 19 ., Therefore , we proceeded to map eQTLs , variance QTLs ( vQTLs ) , CV-QTLs , and Fano-QTLs , and then asked whether we could discover variance effects which could not be explained as effects on mean expression ., We found 235 eQTLs , 5 vQTLs , 0 CV-QTLs , and 0 Fano-QTLs ( FDR 10%; S5 Fig ) ., To validate the eQTLs , we estimated the replication rate against eQTLs discovered in bulk RNA-seq from the same iPSC lines 20 ., We found that 79% of the single cell eQTLs replicate in the matched bulk data ( Fig 2A ) , and 80% of bulk eQTLs replicate in the single cell data ., However , we found 1 , 390 eQTLs ( FDR 10% ) using all of the individuals in the bulk RNA-seq study ( n = 58 ) , and still recovered 1 , 136 eQTLs ( FDR 10% ) after subsampling to n = 53 ., Our results therefore suggest that eQTL discovery in scRNA-seq ( as opposed to replication of previously discovered eQTLs ) loses power compared to equal-sized studies in bulk RNA-seq , likely due to increased experimental noise ., We found 85% of the eQTLs were also discovered as vQTLs ( when restricting to testing only at the eQTL SNP ) , and 100% of vQTLs were discovered as eQTLs ( Fig 2B ) ., We then sought to directly explain away vQTLs as eQTLs by regressing out the mean from the variance ., Treating the residuals from the regression as the phenotype , we recovered zero vQTLs ., These results suggest the significant variance effects detected in this study are all likely to be explained as effects on mean expression ., Our goal in this study was to find QTLs which alter the variance of gene expression independently of altering the mean expression ., Under our model , these QTLs should explain variation in the dispersion parameter across individuals; however , we failed to find dQTLs ., Further , all of the vQTLs we were able to identify could be explained by mean effects ., In contrast , we were able to discover eQTLs , but fewer than expected based on bulk RNA-seq in matched samples ., To understand why we failed to discover dQTLs , and why we discovered fewer eQTLs than expected , we first derived the power function in terms of effect size ( log fold change ) , sample size , noise ratio ( ratio of measurement error variance to phenotypic residual variance ) , and significance level ( Methods ) ., We then sought to estimate the distribution of QTL effect sizes and the typical noise ratio , for both mean expression and dispersion ., To estimate the distribution of QTL effect sizes , we fit a flexible unimodal distribution for the true effect sizes which maximizes the likelihood of the observed effect sizes and standard errors 21 ., Surprisingly , we found that dQTL effects could be larger than eQTL effects ( S6 Fig ) ., For example , we estimate that the 99th percentile eQTL effect size is 0 . 022 , but is 0 . 090 for dQTLs ., Given this result and the power function we derived , there are two possible explanations for why we still failed to find dQTLs: ( 1 ) the noise ratio of dispersion is large ( measurement error reduced power ) , or ( 2 ) the residual variance of dispersion is large ( genetic variation explains little phenotypic variance ) ., To estimate the typical noise ratio , we developed a two-step procedure to estimate the measurement error variance and residual variance per gene ( Methods ) ., Briefly , in our approach we have one measurement error variance per individual , per gene , which equals the sampling variance of our ZINB model ., To estimate each error variance , we used non-parametric bootstrapping ., To estimate the measurement error variance for each gene , we took the median of the estimated measurement error variances across individuals ., To estimate the residual variance for each gene , we fit a flexible unimodal distribution for the true phenotypes which maximizes the likelihood of the observed phenotypes and measurement errors , and estimated the variance of the posterior mean true phenotypes ., Using our approach , we estimated that the typical noise ratio of the dispersion is 2 . 99 , compared to 4 . 18 for the mean ( S7 Fig ) ., This result suggests that we did not fail to find dQTLs only due to measurement error , because the noise ratio was lower for dQTLs than for eQTLs ., As a reference point , a noise ratio equal to 1 has the same impact on power to detect a QTL as cutting the sample size in half , explaining why our study lost power to detect eQTLs ., We found that the typical phenotypic standard deviation of dispersion is 7 . 2 fold larger than that of the mean expression , suggesting we failed to find dQTLs because the effect sizes of dQTLs ( relative to phenotypic standard deviation ) are smaller than the effect sizes of eQTLs ., We finally asked how much power our current study had to detect the 99th percentile dQTL effect size , assuming the typical noise ratio estimated above ., We found that our study had only 0 . 001% power to detect that effect size at Bonferroni–corrected level α = 5 × 10−6 ( Fig 3 ) ., Fixing the typical noise ratio ( a function of the number of cells per individual and sequencing depth ) , we estimate 16 , 015 individuals would be required to achieve 80% power ., As a lower bound ( setting the noise ratio to zero ) , we estimate 4 , 015 individuals would be required regardless of the number of cells per individual ., Overall , our results suggest a much larger study , both in terms of number of individuals and number of cells per individual , would be required to detect the strongest dQTLs in iPSCs ., Individual cells must tolerate both external and internal perturbations arising from the environment or mutations ., It has long been argued that this outcome of robustness is an inherent property of biological systems 22 , and arises from natural selection 23 , 24 ., Robustness is especially critical in the context of cell fate transitions during differentiation 25 ., Other dynamic physiological processes must also be robust , and as a result , loss of robustness is associated with clinically relevant phenotypes and complex genetic disease 26 , 27 ., Cells maintain their identity and other phenotypes despite perturbations because of the robust regulation of key sets of genes 28 ., We hypothesized that QTLs could disrupt the mechanisms underlying robust regulation , and therefore reveal new insights into the genetic regulation of differentiation and disease ., To investigate this hypothesis , we directly observed gene expression variance across multiple individuals using scRNA-seq , and sought to identify QTLs which could alter the variance of gene expression across cells within a single individual , independently of altering the mean expression ., However , we failed to discover such QTLs , and demonstrated that QTLs which are associated with the variance of gene expression can be explained by effects on mean expression ., We found that relative to the phenotypic standard deviation , effects on the dispersion are smaller than effects on the mean , partially explaining why this study failed to find them ., Our results do not rule out genetic effects on variance independent of mean effects , due to limitations of our analysis ., First , our estimated distributions of effect sizes are based on an empirical Bayes estimate of the underlying effect sizes , given the observed effect sizes ., Our results in simulation and observed data suggest the observed effect sizes may be not be accurately estimated given the size of the current study ., Therefore , the empirical Bayes estimate may not accurately reflect the true distribution of effect sizes ., However , we chose to bias the estimation procedure towards putting prior mass on zero , so our estimates of effect sizes are conservative ., Additionally , our estimates may not generalize beyond iPSCs , because the distribution of dispersion effect sizes could vary across cell types and conditions ., Second , we made a strong assumption that latent gene expression is point-Gamma distributed ., In this study , we directly assessed whether or not this was true using a simple statistical diagnostic , and did not find any gross violations of this assumption in the data ., However , it is likely that this assumption will be violated in heterogenous populations of cells ., One possible extension of our method to this case would be to assume there are K homogeneous subpopulations of cells , each described by a ( possibly different ) point-Gamma distribution ., This mixture of ZINB model suggests an expectation-maximization approach where each cell is assigned to a subpopulation , and then the distributions of the subpopulations are re-estimated ., Finally , we took a modular approach to map QTLs in this study: ( 1 ) we estimated parameters for each individual using only the scRNA-seq data , and then ( 2 ) we mapped QTLs using phenotypes derived from the estimated parameters ., An alternative approach would be to include genotype in the count model for the data , and jointly learn the mean , dispersion , proportion of excess zeros , and genetic effect sizes for mean and dispersion ., Such an approach could borrow information across cells with common genotypes to improve power , holding the experiment size fixed ., However , further development will be needed to efficiently fit the models at QTL mapping scale ., We stress that our power calculation is only a rough guideline for designing QTL mapping studies using scRNA-Seq ., Intuitively , some minimum number of cells per individual is required to adequately estimate means and variances ., However , having achieved that lower bound , the most important quantity to maximize is the number of individuals ., In support of this argument , we estimate thousands of individuals would be required to detect a dQTL no matter how many cells were collected per individual ., We based our power calculations on typical values of the noise ratio for the mean expression and dispersion , and chose a conservative significance level ., However , we found considerable variation in the noise ratio across genes , suggesting that our results may not generalize even across genes ., Overall , our results suggest that the technical noise introduced by scRNA-seq greatly reduces the power to discover eQTLs ., Our results also suggest that , for iPSC lines , dramatically larger studies will be required to map both eQTLs and dQTLs from scRNA-seq ., The cell lines used in this study were obtained from the NHGRI Sample Repository for Human Genetic Research at the Coriell Institute for Medical Research ., All samples were collected by the Coriell Institute for Medical with written informed consent and with IRB approval ., We cultured YRI iPSCs 20 in feeder-free conditions for at least ten passages in E8 medium ( Life Technologies ) 29 ., We collected cells using the C1 Single-Cell Auto Prep IFC microfluidic chip ( Fluidigm ) ., We used a balanced block-incomplete design to randomize individuals across chips ., For each chip , we freshly prepared a mixture of cell suspensions from four individuals ., We measured live cell number via trypan blue staining ( ThermoFisher ) , to ensure equal cell numbers across individuals per mixture ., We performed single cell capture and library preparation as previously described using 6 bp Unique Molecular Identifiers 14 ., We pooled the 96 samples on each C1 chip and sequenced them on an Illumina HiSeq 2500 using the TruSeq SBS Kit v3-HS ( FC-401-3002 ) ., We mapped the reads to human genome GRCh37 ( including the ERCC spike-ins ) with Subjunc 30 , deduplicated the UMIs with UMI-tools 31 , and counted molecules per protein-coding gene ( Ensembl 75 ) with featureCounts 32 ., We then matched single cells back to YRI individuals using verifyBamID 33 ., We filtered samples on the following criteria , derived as previously described 14: We filtered genes for QTL mapping on the following criteria: We applied principal component analysis ( PCA ) to the matrix X of log counts per million ( log CPM ) , using the pseudocount proposed in edgeR 34 ., We corrected for gene detection rate by simultaneously regressing out quantiles of gene expression , correcting for sample-specific and gene-specific means , and performing PCA ., Let X = ( x1 , … , xn ) be observed p-vectors , and let ( z1 , … , zn ) be latent k-vectors where k ≪ p ., Then , PCA corresponds to maximum likelihood estimation in the following latent variable model 35:, x i ∼ N ( · ; W z i + μ , σ 2 I ) ( 1 ) In this parameterization , μ denotes a per-coordinate mean ( in our application , per-gene ) ., However , as previously reported 15 , we additionally have to account for the per-sample mean ., Our approach is based on the latent variable model:, x i j ∼ N ( W j z i + q i ′ β j + u i + v j , σ 2 I ) ( 2 ), where u is an n-vector of per-sample means , v is a p-vector of per-gene means , and Q = ( q1 , … , qn ) is a n × k matrix of expression quantiles ., We fit the model as follows: We estimated the squared correlation between each PC and categorical covariates ( batch , C1 chip , individual , well ) by recoding each category as a binary indicator , fitting a multiple linear regression of the PC loadings against the binary indicators , and then estimating the coefficient of determination of the model ., We assume the count data are generated by a zero-inflated negative binomial ( ZINB ) distribution ( S1 Text ) ., Let: Then , we assume:, r i j k∼ Poisson ( · ; R i j exp ( x i j ′ β k ) λ i j k ) ( 6 ), λ i j k∼ π i k δ 0 ( · ) + ( 1 - π i k ) Gamma ( · ; μ i k , ϕ i k ) ( 7 ) Under this model , the mean and variance of gene expression are:, E λ i j k = ( 1 - π i k ) μ i k ( 8 ) V λ i j k = ( 1 - π i k ) μ i k 2 ϕ i k + π i k ( 1 - π i k ) μ i k 2 ( 9 ) Considering just the non-zero component , marginalizing out λ yields the negative binomial ( NB ) log likelihood , weighted by 1 − πik:, l ( · ) = ln ( 1 - π i k ) + r i j k ln ( R i j exp ( x i j ′ β k ) μ i k R i j exp ( x i j ′ β k ) μ i k + ϕ i k - 1 ) + ϕ i k - 1 ln ( ϕ i k - 1 R i j exp ( x i j ′ β k ) μ i k + ϕ i k - 1 ) + ln Γ ( r i j k + ϕ i k - 1 ) - ln Γ ( r i j k + 1 ) - ln Γ ( ϕ i k - 1 ) ( 10 ) Then , marginalizing over the mixture yields the ZINB log likelihood:, ln p ( r i j k ∣ · ) = ln ( π i k + exp ( l ( · ) ) ) if r i j k = 0 ( 11 ) ln p ( r i j k ∣ · ) = l ( · ) otherwise ( 12 ) To estimate the model parameters , we maximized the ZINB log likelihood ., The parameters must satisfy the constraints μik > 0 , ϕik > 0 , 0 ≤ πik ≤ 1 ., To make the problem easier , we re-parameterized in terms of ln μik , ln ϕik , logit ( πik ) and performed unconstrained optimization ., The ZINB log likelihood is nonconvex; therefore , we used a two stage optimization procedure ., In the first stage , we optimized the NB log likelihood with respect to ln μik , ln ϕik , initializing from zero ., In the second stage , we used the NB solution and logit ( πik ) = −8 ( corresponding to a suitably small value of πik ) as the initialization and optimized the ZINB log likelihood ., In both stages , we used batch gradient descent for 30 , 000 iterations with fixed learning rate 10−3 , accelerated by RMSProp 36 ., We implemented the method in Tensorflow 37 ., We defined the size factor of each cell as the total number of molecules detected ( before excluding genes in QC ) ., To correct for technical confounders , we included C1 chip as an observed confounder , recoded as binary indicator variables and centered ., This approach is sufficient to also correct for batch , because in our experimental design , batch is a linear combination of C1 chip ., Intuitively , if there were a batch effect independent of C1 chip , then we could add the batch effect to each chip effect and set the batch effect to 0 ., To assess the goodness of fit of the method , we used a diagnostic test based on the following simple fact: if the data x1 , … , xn are continuous random variables generated from a continuous CDF F , then F ( xi ) ∼ Uniform ( 0 , 1 ) ., Then , to test for goodness of fit of an estimated F ^ to the data x1 , … , xn , we apply the Kolmogorov-Smirnov ( KS ) test to test whether the values F ^ ( x 1 ) , … , F ^ ( x n ) are uniformly distributed ., ( This test is slightly conservative because it uses the data to estimate F ^ ) ., Here , we have to modify this simple procedure to account for the fact that our data are discrete counts , so F is not continuous ., To address this issue , we used randomized quantiles 38: we sample one random value per observation u i ∣ x i ∼ Uniform ( F ^ ( x i - 1 ) , F ^ ( x i ) ) ., These have the property that if xi ∼ F then ui ∼ Uniform ( 0 , 1 ) ., In our model , each observed UMI count xijk comes from a different distribution Fijk , because it depends on the library size which is cell-specific ., We therefore draw u i j k ∣ x i j k ∼ Uniform ( F ^ i j k ( x i j k - 1 ) , F ^ i j k ( x i j k ) ) ., Then , for each individual i and gene k , we apply the KS test to whether the randomized quantiles uijk across cells j are uniformly distributed ., We imputed dosages for 120 Yoruba individuals from the HapMap project ( Phase 3 , hg19 ) as previously described 39 ., We restricted our analysis to 8 , 472 , 478 variants with minor allele frequency at least 0 . 05 ., For each single cell expression phenotype tested , we standardized and quantile-normalized the phenotype matrix to a standard normal as previously described 40 ., We called QTLs within 100 kilobases of the transcription start site of each gene and controlled the gene-level false discovery rate using QTLtools 41 ., We included principal components ( PCs ) of the normalized expression matrix as covariates for QTL mapping , and selected the number of PCs for each phenotype by greedily searching for the number of PCs which maximized the number of QTLs discovered on even chromosomes only at FDR 10% ., We did not include genotype PCs as covariates ., We additionally recalled eQTLs in the matched bulk RNA-seq data 20 using the re-processed dosage matrix ., We performed replication testing by taking each SNP-gene pair from the discovery cohort , and testing that pair in the replication cohort ., We defined a hit as replicating if it passed the Benjamini–Hochberg procedure at level 10% ( restricted to the set of SNP-gene pairs tested ) and had the same effect size direction ., For individual i and gene k , we assume the generative model:, y i k= x i b + e i k ( 13 ) y ˜ i k= y i k + e ˜ i k ( 14 ), where y ˜ i k is the observed phenotype , yik is the true phenotype , xi is the genotype at the SNP of interest , e ˜ i k ∼ N ( 0 , σ m 2 ) , and e i k ∼ N ( 0 , σ r 2 ) ., To perform QTL mapping , we fit a working model which ignores measurement error:, y ˜ i k = x i β + ϵ i k ( 15 ), where ϵik ∼ N ( 0 , σ2 ) ., From this model , we estimate β ^ ., Assuming V x = 1 , we have σ 2 = σ r 2 + σ m 2 and:, β ^ ∼ N ( b , σ r 2 + σ m 2 n ) ( 16 ), where n is the number of individuals ., Under the working model , the power function is:, Pow ( · ) = Φ ( Φ - 1 ( α 2 ) + b SE ( β ^ ) ) ( 17 ), where α denotes the significance level , SE ( ⋅ ) denotes standard error , and Φ ( ⋅ ) denotes the standard Gaussian CDF ., Under the assumed generative model , the power function equals:, Pow ( λ , n , δ , α ) = Φ ( Φ - 1 ( α 2 ) + λ n 1 + δ ) ( 18 ), where λ = b/σr , and δ = σ m 2 / σ r 2 ., Parameterized in terms of δ , the power function implies useful reference points; for example , δ = 1 is equivalent to cutting the sample size in half ., To determine the effect size b , we estimate the distribution of true effect sizes b given observed effect sizes β ^ j and associated standard errors s ^ j ., We assume the hierarchical model:, β ^ j ∣ b j , s ^ j∼ N ( b j , s ^ j 2 ) ( 19 ) b j ∣ s ^ j∼ g ( · ) ( 20 ), where g is a unimodal mixture of Gaussians ., We estimate g using adaptive shrinkage ( ash ) 21 ., We took b to be the 99th percentile of the fitted distribution ., Although we assumed a single measurement error variance σ m 2 , we actually have measurement errors for each individual and gene σ m i k 2 ., To estimate σ m i k 2 , we used non-parametric bootstrapping ., For each individual and gene , we resampled the counts ( matched with the library size and technical confounders ) with replacement , and refit the ZINB model ., To reduce computational burden , we restricted our analysis to 200 randomly chosen genes , warm-started the optimization from the optimal parameters for the original data , and ran gradient descent for 30 , 000 iterations ., To estimate the typical noise ratio δ , we estimate a measurement error variance per gene σ m k 2 and a residual variance per gene σ r k 2 ., We take σ ^ m k 2 = median ( σ m i k 2 ) ., To estimate σ r k 2 , we solve a deconvolution problem 42:, y ˜ i k ∣ y i k , σ ^ m i k 2∼ N ( y i k , σ ^ m i k 2 ) ( 21 ) y i k ∣ σ ^ m i k 2∼ g ( · ) ( 22 ), where g is a unimodal mixture of uniforms , estimated using ash ., To fit the model , we centered the y ˜ i k for each gene k , concatenated them across genes , and assumed a common prior ., Then , the required estimates are:, σ ^ r k 2= V ^ E y i k ∣ · ( 23 ) δ ^= median ( σ ^ m k 2 σ ^ r k 2 ) ( 24 ) λ= b median ( σ ^ r k 2 ) ( 25 ), where V ^ denotes sample variance . | Introduction, Results, Discussion, Materials and methods | Quantification of gene expression levels at the single cell level has revealed that gene expression can vary substantially even across a population of homogeneous cells ., However , it is currently unclear what genomic features control variation in gene expression levels , and whether common genetic variants may impact gene expression variation ., Here , we take a genome-wide approach to identify expression variance quantitative trait loci ( vQTLs ) ., To this end , we generated single cell RNA-seq ( scRNA-seq ) data from induced pluripotent stem cells ( iPSCs ) derived from 53 Yoruba individuals ., We collected data for a median of 95 cells per individual and a total of 5 , 447 single cells , and identified 235 mean expression QTLs ( eQTLs ) at 10% FDR , of which 79% replicate in bulk RNA-seq data from the same individuals ., We further identified 5 vQTLs at 10% FDR , but demonstrate that these can also be explained as effects on mean expression ., Our study suggests that dispersion QTLs ( dQTLs ) which could alter the variance of expression independently of the mean can have larger fold changes , but explain less phenotypic variance than eQTLs ., We estimate 4 , 015 individuals as a lower bound to achieve 80% power to detect the strongest dQTLs in iPSCs ., These results will guide the design of future studies on understanding the genetic control of gene expression variance . | Common genetic variation can alter the level of average gene expression in human tissues , and through changes in gene expression have downstream consequences on cell function , human development , and human disease ., However , human tissues are composed of many cells , each with its own level of gene expression ., With advances in single cell sequencing technologies , we can now go beyond simply measuring the average level of gene expression in a tissue sample and directly measure cell-to-cell variance in gene expression ., We hypothesized that genetic variation could also alter gene expression variance , potentially revealing new insights into human development and disease ., To test this hypothesis , we used single cell RNA sequencing to directly measure gene expression variance in multiple individuals , and then associated the gene expression variance with genetic variation in those same individuals ., Our results suggest that effects on gene expression variance are smaller than effects on mean expression , relative to how much the phenotypes vary between individuals , and will require much larger studies than previously thought to detect . | null | null |
journal.pntd.0000568 | 2,009 | Human Probing Behavior of Aedes aegypti when Infected with a Life-Shortening Strain of Wolbachia | Insect transmitted diseases such as malaria and dengue occur in more than 100 countries worldwide , placing at risk around half the worlds population ., The disease burden is high with more than 500 million cases each year ., Despite various vector control measures , there continues to be emergence and resurgence of these diseases 1 ., Aedes aegypti is the main vector of dengue fever causing millions of cases and thousands of deaths each year ., Vector control is the only method for dengue and dengue hemorrhagic fever ( DHF ) prevention , however current strategies are failing to prevent the increasing global incidence of dengue fever 2 ., The development of practical alternative strategies to control dengue , which could be used in conjunction with current measures , is much needed ., Recently our group reported the successful stable infection of A . aegypti with the wMelPop Wolbachia strain that reduces insect lifespan 3 ., Wolbachia is an inherited bacterium able to manipulate the insect hosts reproductive biology 4 in a manner that promotes its rapid spread through insect populations 5 ., Releasing Wolbachia-infected mosquitoes that could initiate an invasion of Wolbachia into a wild mosquito population 6 and that resulted in reduced lifespan of wild mosquitoes could theoretically greatly reduce transmission of dengue virus ., This is because only old mosquitoes transmit the virus 3 , 7 , 8 ., In addition to lifespan reduction , we have recently shown that the wMelPop infection substantially reduces dengue load in A . aegypti mosquitoes 9 and reduces their ability to successfully obtain blood meals as they age 10 ., Mosquitoes rely on chemical and physical cues ( e . g . carbon dioxide , body odors , air movement and heat ) to locate suitable hosts for feeding 11 , 12 and this complex set of activities is known as host-seeking behavior 13 , 14 ., Once the host is located , the mosquito must quickly obtain blood to avoid any host defensive behavior 15 , 16 ., Behavior during feeding can then be divided into the stages of pre-probing ( foraging ) 17 and probing ( feeding ) activities 18 ., After finding a suitable blood vessel and thrusting its stylet into the host skin 19 , the saliva plays an important role in preventing blood clotting , through the anti-platelet aggregation activity of the enzyme apyrase 20 , 21 and other antihemostatic and anti-inflammatory compounds 19 ., Insect vector competence for the transmission of viruses and parasites is dependent upon the successful execution of these upstream steps in the process ., Here we report the results of an examination of the effects of wMelPop infection on pre-probing behavior , probing behavior , saliva production , and apyrase content of saliva of A . aegypti ., The goals of the study were to identify possible mechanisms for the insects reduced ability to obtain a blood meal with age and further evaluate the capacity for Wolbachia infection to reduce vector competence ., This study was conducted according to the principles expressed in the Declaration of Helsinki ., The study was approved by the Medical Research Ethics Committee at the University of Queensland ( Project #2007001379 ) ., Volunteers were made aware of the risks of bloodfeeding ( allergy and discomfort ) and the plans to analyze and publish all data prior to providing written consent to participate in the study ., Aedes aegypti mosquitoes , wMelPop infected ( PGYP1 ) and its Tetracycline-cured counterpart ( PGYP1 . tet ) 3 , were kept in a controlled environment insectary at 25 °C and 80% RH ., Larvae were maintained with fish food pellets ( Tetramin , Tetra ) and adults were offered 10% sucrose solution , ad libitum ., Adult females were fed on human blood ( UQ human ethics approval 2007001379 ) for egg production and eggs were dried for at least 96 h prior to hatching ., Fertilized and non-blood fed females of different ages ( 5 , 15 , 26 and 35 days old ) were used in all behavior experiments ., Sucrose solutions were removed from cages on the night before the experiments ., Forty females were used per age and per infectious status ., Single mosquitoes were transferred to a transparent Perspex cage ( 25 cm3 ) and filmed through a digital camera with 6mm Microlens ( IEEE-1394 , Point Grey Research ) mounted on a tripod ., Mosquitoes were given about five minutes to settle within the cage before a human gloved-hand was inserted into the cage ., A window of about 15 cm2 was cut of the upper part of the latex glove in order to delineate the probing field ., Movies were recorded ( QuickTime Player ) for a maximum of 10 minutes or until blood was seen within the mosquito midgut and sub-sequentially watched for time calculations ., Two electronic timers were used , one for recording pre-probing time and the second for probing time ., Pre-probing time was defined and the time since the mosquito has landed on the bare hand area until the insertion of mouthparts into the human skin ., Probing time is defined as the initial insertion of insect mouthparts until blood can be seen within the mosquito midgut through the abdominal pleura 21 ., Timing stopped when mosquitoes left the bare hand area or withdrew their mouthparts before taking blood and began again when the mosquito came back or after subsequent stylet penetration ., If blood was not found by the end of 10 minutes , we defined this case as unsuccessful probing and it was measured as a proportion ., Movies were also used to visualize additional abnormal phenotypes as the jittering action of mosquito body while landed on top of the human hand , and named “shaky” ., Furthermore , the inability of mosquitoes to insert they mouthparts due to a bendy proboscis was also analyzed ., The bendy phenotype was recently described by Turley et al 10 ., Mosquitoes of different ages ( 5 , 26 and 35-days-old ) and infectious status were starved overnight ( without sucrose solution or water ) ., On the following morning mosquitoes were briefly anesthetized with CO2 and placed on a glass plate over ice ., Wings and legs were removed with forceps and their proboscis introduced into a 1cm piece of polypropylene tubing ( 0 . 61×0 . 28mm , Microtube Extrusions , NSW , Australia ) ( modified from 21 ) ., Females were allowed to salivate for 5 minutes and then the diameter of the saliva droplets was measured through an ocular micrometer at 40× magnification ., Volumes were calculated via the sphere formula 22 ., Saliva was then collected into 20 µL of 0 . 05mM Tris-HCl pH 7 . 5 by attaching the needle of a 10 µL Hamilton syringe and rinsing the tubing content a few times ., Samples were centrifuged at 14 , 000g for 2 minutes and kept frozen ( −80°C ) in 20 µL of 0 . 05mM Tris-HCl , pH 7 . 5 for enzymatic assay ( see below ) ., Saliva samples ( 8 µL ) were transferred , in duplicates , into individual wells of a plastic 96-well ELISA plate ( NUNC Maxisorp ) ., For the blank , 8 µL of the 0 . 05mM Tris buffer was added to the wells ., To each well , 100 µL was added of a mixture containing 100mM NaCl , 50mM Tris–HCl ( pH 8 . 95 ) , 5mM CaCl2 , 2mM ATP and 20 mM B-Mercapthanol ., The plate was incubated at 37°C for 10 min and then the reaction was immediately stopped , by adding 25 uL of acid molybdate solution ( 1 . 25% ammonium molybdate in 2 . 5mM H2SO4 ) ., Immediately after termination of the reaction , 2 µL of a reducing solution ( 0 . 11mM NaHSO3 , 0 . 09mM Na2SO3 and 8mM 1-amino-2-naphthol-4-sulphonic acid ) was added to each well and the plate was incubated at 37°C for 20 min 22 ., Plates were read at a FLUOstar OPTIMA ELISA plate reader ( BMG Technologies ) at 660nm ., Readings were quantified by comparison with an inorganic phosphate standard curve ( 1 , 0 . 5 , 0 . 25 , 0 . 125 , 0 . 06125 , 0 . 03125 , 0 . 015625 mM of sodium phosphate ) ., Wolbachia infection was confirmed through PCR to detect both mosquito ( apyrase gene: ApyF: 5′-TTTCGACGGAAGAGCTGAAT-3′ and ApyR: 5′-TCCGTTGGTATCCTCGTTTC-3′ ) and Wolbachia ( IS5-F: 5′-CTGAAATTTTAGTACGGGGTAAAG-3′ and IS5-R: 5′- CAAGCATATTCCCTCTTTAAC-3′ ) sequences ., Saliva screening to check the presence of Wolbachia was also done via PCR ( with IS5 primers ) using saliva samples of infected and non-infected mosquitoes ., Mosquito sequences in this case were detected with primers for the ribosomal protein gene RpS17 23 ., In all cases , general linear models were employed to examine the effects of the variables age and infection status and their interaction with one another ., Models demonstrating significance for the variable infection status were then followed by individual t-tests examining the differences between infected and uninfected mosquitoes for each age class ., The proportion of infected and uninfected mosquitoes that obtained blood meals were examined using Mann-Whitney U tests instead of linear models , given the deviation of the data from normality , and were based on four populations of mosquitoes ., Chi-square 2×2 contingency tests were employed to examine the relationship between observed behavioral traits and lack of feeding success ., The correlation between these traits was quantified using a cox-proportional hazards model for age , with the behavioral traits and lack of blood meal success covariates ., All statistical analyses were carried out in STATISTICA v8 ( StatSoft , Tulsa , OK ) ., We measured the time mosquitoes spent from first contact with a human volunteer until the insertion of the insects mouthparts as a measure of pre-probing time ., All feeding trials were carried out with individual mosquitoes , which had been starved prior to the assay , at four adult ages ( 5 , 15 , 26 and 35-days-old ) ., Mosquitoes that never successfully achieved a blood meal were excluded from this analysis ., Overall both age ( df\u200a=\u200a3 , F\u200a=\u200a13 . 73 , P<0 . 0001 ) and infection status ( df\u200a=\u200a1 , F\u200a=\u200a23 . 18 , P<0 . 0001 ) had a significant effect on the length of pre-probing time ., On average infected mosquitoes spend more time pre-probing especially as they age ( Fig . 1 ) ., This change with age is clearly exhibited by a significant interaction between the variables age and infection status ( df\u200a=\u200a3 , F\u200a=\u200a8 . 11 , P<0 . 0001 ) ., At five days of age infected and uninfected mosquitoes do not differ in their pre-probing time ( df\u200a=\u200a78 , t\u200a=\u200a0 . 63 , P\u200a=\u200a0 . 52 ) , which lasted on average 11 seconds ., Uninfected mosquitoes maintained the same foraging time as they aged , while wMelPop insects exhibited a steady and significant increase ( 15d: df\u200a=\u200a75 , t\u200a=\u200a−3 . 37 , P\u200a=\u200a0 . 0012; 26d: df\u200a=\u200a63 , t\u200a=\u200a−4 . 17 , P\u200a=\u200a0 . 014; 35d: df\u200a=\u200a48 , t\u200a=\u200a−2 . 25 , P\u200a=\u200a0 . 0034 ) , reaching a mean length of 45 sec by 35 days of age ( Fig . 1 ) ., In the same feeding trials described above , the length of time between insertion of mouthparts and the first visible sign of blood in the abdominal pleura 21 was recorded as probing time for the mosquitoes ., As with pre-probing time , the variables of age ( df\u200a=\u200a3 , F\u200a=\u200a11 . 36 , P<0 . 0001 ) , infection status ( df\u200a=\u200a1 , F\u200a=\u200a29 . 46 , P<0 . 0001 ) and the interaction ( df\u200a=\u200a3 , F\u200a=\u200a10 . 56 , P<0 . 0001 ) between these two variables were highly significant ., Infected and uninfected mosquitoes did not differ in their probing time ( ∼33 sec ) at 5 ( df\u200a=\u200a78 , t\u200a=\u200a−0 . 46 , P\u200a=\u200a0 . 64 ) and 15 ( df\u200a=\u200a75 , t\u200a=\u200a1 . 43 , P\u200a=\u200a0 . 15 ) days of age ( Fig . 2 ) ., In contrast , infected mosquitoes at 26 ( df\u200a=\u200a63 , t\u200a=\u200a−3 . 76 , P<0 . 001 ) and 35 ( df\u200a=\u200a48 , t\u200a=\u200a−4 . 06 , P<0 . 001 ) days of age took significantly longer during probing , exhibiting up to a seven-fold increase in their probing time relative to uninfected mosquitoes ( Fig . 2 ) ., In the assays detailed above we then compared the ability of infected and uninfected mosquitoes to obtain blood meals ( Fig . 3 ) using Mann-Whitney U tests ., At 5 ( Z\u200a=\u200a0 , P\u200a=\u200a1 ) and 15 ( Z\u200a=\u200a0 , P\u200a=\u200a1 ) days of age infected and uninfected mosquitoes did not differ in their success ., At 26 ( Z\u200a=\u200a−2 . 39 , P\u200a=\u200a0 . 020 ) and 35 ( Z\u200a=\u200a−2 . 39 , P\u200a=\u200a0 . 020 ) days of age infected mosquitoes were less successful at obtaining blood meals in comparison to their uninfected counterparts ., Only Wolbachia infected mosquitoes refused to land on the hand during the assay or landed but did not take a blood meal ., Percentages of individuals that did not land on the human were: 2 . 5% at 15 d; 0% at 26 d and 7 . 5% at 35d ., Percentages that landed but that did not feed were 5% at 15 d; 37 . 5% at 26d and 67 . 5% at 35 d ., It is important to note that as infected mosquitoes aged , the frequency of events where they pierced the skin did not increase despite failed attempts at feeding ( Fig . 4 ) ., A general linear model revealed that age ( df\u200a=\u200a3 , F\u200a=\u200a20 . 47 , P<0 . 0001 ) , infection ( df\u200a=\u200a3 , F\u200a=\u200a29 . 12 , P<0 . 0001 ) and age X infection ( df\u200a=\u200a3 , F\u200a=\u200a27 . 18 , P<0 . 0001 ) were significant determinants of the number of probings ., Subsequent t-tests comparing the number of probings between infected and uninfected mosquitoes at each of the age points ( data not shown ) , however , demonstrated that only 35 day old ( df\u200a=\u200a1 , t\u200a=\u200a−8 . 44 , P<0 . 0001 ) mosquitoes differed ., In this case , uninfected females probed more on average per session ( 1 . 05±0 . 05 ) than wMelPop infected mosquitoes ( 0 . 3±0 . 073 ) ., This is due to other behaviors , which impaired the infected mosquitoes to feed ( see below ) ., We recently have reported the appearance of a “bendy” proboscis in association with wMelPop , which was the inability of the mosquito to properly orient its mouthparts and insert the stylet into the skin 10 ., Here we quantified the occurrence of this trait ., The bendy proboscis was never observed in any of the uninfected mosquitoes in this study , nor was it present in a small cohort of very old mosquitoes ( ∼90 days ) we examined in a second pilot study ., The phenotype was also not present in 5 day-old infected mosquitoes ., The trait first appeared at a low level ( 2 . 5% ) in 15 day-old wMelPop infected mosquitoes and rose to a frequency of 65% by 35 days of age ( Fig . 5 ) ., Another phenotype observed , although in lower frequencies , was the jittering action of the insect body ( named here as “shaky” ) when the mosquito was sitting on top of the human hand ( Fig . 5 , Video S1 ) ., The association between each of these traits and lack of success in blood meal acquisition was explored using 2×2 contingency tests in each of the age classes where the trait was expressed ., There was a significant association between the failure to obtain a blood meal and both the bendy phenotype ( 26d: df\u200a=\u200a1 , χ2\u200a=\u200a14 . 1 , P\u200a=\u200a0 . 0002; 35d: df\u200a=\u200a1 , χ2\u200a=\u200a11 . 8 , P\u200a=\u200a0 . 0006 ) and the shaky phenotype ( 35d: df\u200a=\u200a1 , χ2\u200a=\u200a4 . 2 , P\u200a=\u200a0 . 038 ) ., Using survival analysis we obtained estimates of the correlation between lack of feeding success and the bendy phenotype ( 0 . 63 ) and the shaky phenotype ( 0 . 19 ) ., These correlations reveal the presence of a relationship between the traits and success in feeding , but do not completely explain lack of success ., There are mosquitoes in the older age classes that fail to feed and that are not shaky or bendy ., To discard any possibility that this other abnormal phenotypes were due to the lack of blood feeding , which could have physiologically compromised the mosquitoes we also blood fed females of both groups when they were 3 to 5-days-old and then after 38 days evaluated their feeding behavior ., None of the wMelPop mosquitoes were able to feed and all presented the bendy proboscis , although all the tetracycline-treated mosquitoes successfully imbibed blood ( data not shown ) ., To check whether the probing behavior and the additional phenotypes we observed were due to differences in saliva volume and salivary gland apyrase activity we measured both traits in infected and uninfected mosquitoes at three adult ages ., Apyrase activity ( Fig . 6A ) did not differ in infected and uninfected mosquitoes regardless of age ( df\u200a=\u200a1 , F\u200a=\u200a0 . 44 , P\u200a=\u200a0 . 51 ) ., Infection status ( df\u200a=\u200a1 , F\u200a=\u200a11 . 99 , P<0 . 01 ) and age ( df\u200a=\u200a2 , F\u200a=\u200a14 . 54 , P<0 . 0001 ) , however , were determinants of saliva volume ( Fig . 6B ) and on average infected mosquitoes produced less saliva ., When saliva volumes of infected and uninfected mosquitoes were compared to each other for each age class , only the 26 days old mosquitoes were significantly different ( df\u200a=\u200a1 , t\u200a=\u200a−2 . 9 , P<0 . 01 ) ., During collection , 80–100% of the mosquitoes produced saliva droplets ., Infection status was not a predictor of whether saliva was produced or not ( data not shown ) ., To begin to dissect the functional role of Wolbachia in these feeding effects we sought to determine the presence of the bacterium in both the saliva and salivary glands ., PCR amplification of the Wolbachia wsp gene or mosquito apyrase has shown only the presence of Wolbachia in salivary glands , but not in saliva ( see Figure S1 ) ., The transposable element IS5 , present in at least 13 copies within the bacteria genome 24 , was also used in extra samples as a very sensitive PCR target ( N\u200a=\u200a16 of each group ) but no amplification was obtained ( data not shown ) ., These results are supported by the size of the intracellular Wolbachia ( around 1µm in diameter ) 4 and the diameter of mosquito salivary ducts ( also about 1 µm ) 25 , which indicate that even if Wolbachia was able to be present in the secreted salivary fluid it would be unlikely to be able to freely move through the salivary ducts ., The results presented here indicate that infection with wMelPop alters the blood-feeding behavior of female A . aegypti mosquitoes in a manner that intensifies with increasing mosquito age ., Older infected mosquitoes spend more time pre-probing and probing , commonly exhibit evidence of shaking and a bendy proboscis , produce less saliva and overall are less successful in obtaining blood meals ., None of our experiments revealed any evidence of Wolbachia in the saliva or Wolbachia-associated changes in salivary apyrase activity ., Female mosquitoes must balance the feeding time required to obtain a sufficient blood meal against the risk of death from human defensive behavior upon their detection 26 ., On the other hand , parasites or viruses that are transmitted through the insect saliva benefit from feeding times that enhance transmission but possibly cause a higher risk for the mosquito ., Parasite manipulation of insect feeding behavior often includes increases in the insects probing time or number of probings ., This has been seen in mosquitoes infected with malaria parasites 27 and viruses 28 , and in trypanosome-infected triatomine bugs 29 ., As Wolbachia is not transmitted via saliva , the increased length of time infected mosquitoes are taking to feed could be advantageous only if blood-meal sizes acquired were greater and subsequent fecundity rates higher ., Our group recently demonstrated evidence to the contrary , with infected females taking smaller blood meals on average 10 ., This in conjunction with the data reported here , with overall reduced rates of feeding in combination with physiological changes like the observed “bendy” and “shaky” traits , suggest that the feeding biology is not likely to be a Wolbachia parasite “adaptation . ”, In the absence of evidence for a specific Wolbachia manipulation of the host , blood-feeding effects are likely explained as a by-product of infection or as the direct result of pathogenesis on host cells or tissues ., Wolbachia are found in many different insect tissues , especially in nervous tissue , muscles and reproductive systems 9 , 30 ., While in Drosophila the deleterious effects of wMelPop are related to age and bacterial density 4 , the relationship between bacterial density and pathogenicity has yet to be empirically determined for wMelPop-infected A . aegypti ., The decreasing ability to feed with increasing age in combination with the increased prevalence of shaking behavior and the bendy proboscis could be explained by direct damage on host cells or tissues , resulting from Wolbachia over-replication ., Interestingly , failure to obtain a blood meal cannot be completely explained by these two traits ., Alternatively , Wolbachia infection may be having direct and targeted effects on neuro-regulators , which would influence their ability to forage for blood and actually feed ., The bendy proboscis and shaky phenotypes may represent very extreme endpoints in the process ., The wMelPop strain of Wolbachia was initially targeted for biocontrol development given its life-shortening capacity 7 ., More recently , the presence of wMelPop has been shown to block the accumulation of dengue virus in A . aegypti 9 ., The pattern of the reduced blood-feeding success seen here , like life-shortening leaves the young breeding individuals unaffected while targeting older members of the population ., While the dengue transmission rates of wMelPop-infected mosquitoes are yet to be estimated , initial concerns that infected mosquitoes , failing to successfully feed , might actually bite more and hence possibly enhance viral transmission appear unfounded ., While probing frequency decreased with age in Wolbachia infected individuals probing time increased ., This was compensated by a reduction in saliva produced by Wolbachia infected individuals , which should counteract any potential increases to pathogen transmission risk in Wolbachia infected individuals ., In conclusion , the presence of Wolbachia bacterium in A . aegypti mosquitoes significantly reduces their feeding success in an age dependent manner that is more likely a by-product of virulence than a direct manipulation of host behavior ., These effects on feeding could substantially improve the efficacy of wMelPop biocontrol strategies in combination with the traits of life-shortening and viral blocking . | Introduction, Methods, Results, Discussion | Mosquitoes are vectors of many serious pathogens in tropical and sub-tropical countries ., Current control strategies almost entirely rely upon insecticides , which increasingly face the problems of high cost , increasing mosquito resistance and negative effects on non-target organisms ., Alternative strategies include the proposed use of inherited life-shortening agents , such as the Wolbachia bacterium ., By shortening mosquito vector lifespan , Wolbachia could potentially reduce the vectorial capacity of mosquito populations ., We have recently been able to stably transinfect Aedes aegypti mosquitoes with the life-shortening Wolbachia strain wMelPop , and are assessing various aspects of its interaction with the mosquito host to determine its likely impact on pathogen transmission as well as its potential ability to invade A . aegypti populations ., Here we have examined the probing behavior of Wolbachia-infected mosquitoes in an attempt to understand both the broader impact of Wolbachia infection on mosquito biology and , in particular , vectorial capacity ., The probing behavior of wMelPop-infected mosquitoes at four adult ages was examined and compared to uninfected controls during video-recorded feeding trials on a human hand ., Wolbachia-positive insects , from 15 days of age , showed a drastic increase in the time spent pre-probing and probing relative to uninfected controls ., Two other important features for blood feeding , saliva volume and apyrase content of saliva , were also studied ., As A . aegypti infected with wMelPop age , they show increasing difficulty in completing the process of blood feeding effectively and efficiently ., Wolbachia-infected mosquitoes on average produced smaller volumes of saliva that still contained the same amount of apyrase activity as uninfected mosquitoes ., These effects on blood feeding behavior may reduce vectorial capacity and point to underlying physiological changes in Wolbachia-infected mosquitoes . | Mosquitoes transmit diseases when they are actively searching for a source of blood ., This so called probing behavior comprises the “searching” time , the beginning of the feeding process until the first sign of blood can be seen within the insect body ., The manipulation of this behavior can have important consequences for the mosquitos ability to transmit pathogens , such as dengue virus or Plasmodium ., In this study we examined the probing behavior of the main vector of dengue viruses , Aedes aegypti , when infected with an intracellular bacterium , Wolbachia pipientis ., This bacterium alters the probing behavior of older mosquitoes such that they take longer to find a feeding site and longer to imbibe blood , which may make them more susceptible to human defense responses ., The bacterium appears to reduce mosquito feeding success by preventing the mosquito from successfully inserting its stylet into human skin ., The old age onset of reduced mosquito feeding success due to Wolbachia could selectively promote a reduction in dengue transmission . | ecology/behavioral ecology | null |
journal.pcbi.1004914 | 2,016 | Detecting Mild Traumatic Brain Injury Using Resting State Magnetoencephalographic Connectivity | Detection of mild traumatic brain injury ( mTBI ) using neuroimaging remains a challenge , as no abnormalities are typically apparent using routine MRI 1 , 2 ., Accordingly , diagnosis is usually a clinical judgement based on self-report measures and behavioural assessments ., Despite the lack of apparent injury on conventional clinical scans , many patients with mTBI suffer post-concussive symptoms ( PCS ) ., Although such symptoms typically resolve within a few months , a subset of individuals continue to experience long-term cognitive and behavioural impairments 3–5 , underscoring the need for quantitative and objective methods for detecting and determining the severity of mTBI ., The presence of lingering PCS indicates the presence of subtle brain injuries , with significant functional consequences that cannot be detected using current clinical techniques; there is a need to develop new imaging approaches for the detection of mTBI using quantitative and objective evidence ., Recent advances in magnetoencephalographic ( MEG ) imaging indicate that identification of mTBI is possible through detection of excessive slow-wave activity 6 and that this approach can localize the foci of the damage 7 ., MTBI is associated with altered white matter microstructure as indicated by diffusion tensor imaging ( DTI ) , in agreement with the view that mTBI results in axonal injury 8 ., The focal excessive MEG slow wave activity has been shown to be related to the location of white matter injury , consistent with the supposition that oscillatory slowing can occur from deafferentation 9 ., Disruption of inter-regional oscillatory synchrony in mTBI has been reported using EEG 10 ., Oscillatory synchrony among brain areas is understood to play a vital role in network connectivity supporting cognition and behaviour 11 , 12 , and the expression of such neurophysiological network connectivity at rest relates to the intrinsic organization of brain activity pertinent for brain function and its dysfunction in clinical populations 13 ., Converging evidence now indicates that traumatic brain injury is associated with diffuse axonal injury , which disrupts intrinsic functional network connectivity , thereby contributing to associated cognitive sequelae 14 ., Machine learning approaches have been successfully combined with imaging of intrinsic functional brain connectivity during resting state to accurately classify single individuals 15 , 16 ., Moreover , electrophysiological recordings from subjects have also been shown to be effective for accurately determining group membership of individuals 17 , 18 ., We used MEG to investigate alterations in resting state oscillatory network synchrony in adults with mTBI , and investigated the hypothesis that machine learning algorithms could accurately detect mTBI in individual subjects ., Frequency and source resolved imaging of resting MEG network synchrony was able to accurately detect whether individual participants had been diagnosed with mTBI or not ., Fig 1 shows the predictive power of phase synchrony measured at 30 specific frequencies points covering the range between 1Hz and 75Hz ., Specificity remains stable around 80% and fluctuates slightly ( 76–83% ) within a relatively wide range of frequencies: from 3Hz to 50Hz ., Conversely , sensitivity and hence total accuracy have a local maximum around 8-13Hz ( α rhythms ) , reaching 80% ., Fig 2A provides a more aggregated view of the results shown in Fig 1 ., Specifically , prediction accuracy is given as functions of frequency bands , each including the features from several frequency points ( wavelets ) ., Fig 2B illustrates the same prediction accuracies with respect to the random chance prediction , wherein the distributions of accuracies were generated by shuffling the labels ( mTBI or not ) across subjects , and repeating the same procedure 500 times ., With an accuracy of 80% ( p < 0 . 01 ) , inter-regional resting state phase synchrony in the α band carries the most discriminative information for inferring the presence or absence of mTBI within a single individual ., Accordingly , the remainder of the results presented in this section pertain to phase synchrony in the α band , using features that individually provided the highest separability between mTBI and controls under the ROC criterion ., The list of ranked features reflects an estimate of how valuable a given feature was found to be for classification ., We can choose the best number of features , i . e . the number that maximizes prediction accuracy ., In this case , the dimensionality of the feature space will correspond to the number of source-pairings within the alpha range ., Note that in this study the features were ranked using training data at each round of leave-one-out cross-validation ., While the number of features to keep was set a priori , the best features themselves were determined within each round of cross-validation ., Fig 3 shows accuracy values as functions of the number of best features selected for classification analysis with cross-validation ., As can be seen from Fig 3C , classification accuracy can be improved with a proper threshold on the number of variables k with two peaks around k = 8−15 and k = 30−35 ., For example , for k = 33 accuracy is 88% , with 90% specificity and 85% sensitivity ( all p < 0 . 01 ) ., To investigate potential relations between SVM classification and symptom severity obtained from the concussion assessment tool ( SCAT2 ) , we quantified the distance to the decision boundary for each subject , and correlated these values with clinical scores for participants within the mTBI group ., Note that the larger the distance that an individual is from the decision boundary , the higher our confidence that a subject with mTBI is classified correctly as mTBI ., Similar to the procedure shown in Fig 4 , for each subset of features k = 1 , … , 100 , we computed the distances to the decision boundary for mTBI patients , and correlated these distances with the severity and symptom scores , shown in Fig 4A and 4B , respectively ., Two scatter plots with superimposed least-squares regression lines illustrate relations between these variables at two peaks , k = 11 for severity ( Fig 4C ) , and k = 33 for symptoms ( Fig 4D ) ., Note that negative distances at the scatter plots reflect cases of misclassification , when the learning function F ( x ) projects the feature vectors x of mTBI subjects to other side of the optimal hyperplane , corresponding to controls ., Moreover , the confidence of classifying a subject as mTBI positively correlated with the self-reported severity scores ( Fig 4A ) , reaching a local maximum ( r = 0 . 54 , p < 0 . 05 ) at k = 11 ., It also correlated positively with the symptoms scores with a peak of r = 0 . 34 ( p-value< 0 . 10 ) around k = 33 ., Finally , Fig 5 illustrates a distribution of the connections extracted from the pool of the best k = 33 features in the α band ., It plots connections within a transparent template of the brain in the MNI space , using the BrainNet Viewer 19 ., The width of the connections represents the weights are between −1 and 0 , where being close to −1 implies a robust contribution of a specific connection to classification , and zero means no contribution ., Specifically , for each wavelet frequency within the α range and each round of cross-validation ( m = 1 , … , 41 ) , we assigned −1 to a connection if this feature survived the threshold and participated in classification , otherwise it was 0 , subsequently averaging across subjects and wavelets frequencies ., The ability to predict evidence of injury of a subject is largely based on synchrony between frontal and parietal/temporal sites , located mainly in the left hemisphere ., We also employed PLS to characterize and test the statistical reliability of differences in resting state network synchrony between adults with and without mTBI ., This analysis revealed the existence of one significant latent variable ( p = 0 . 002 ) which indicated alterations of resting MEG network synchrony in mTBI ( Fig 6A ) ., The overall distribution of all the bootstrap ratio values , each associated with a pair-wise connection between the sources and frequencies , is shown in Fig 6B ., As can be seen , there are relatively large positive and negative bootstrap ratio values , which reflect phase-locking and phase scattering effects , respectively , in controls with respect to mTBI ., The difference between increased and decreased phase locking is broken down further in Fig 6C and 6D , which shows how the strength of these effects varies across frequencies ., Specifically , we identified the 1% tails , cut off by the 0 . 01− and 0 . 99-quantiles of the overall distribution of the bootstrap ratio values in Fig 6B ., At each frequency , the number of connections with the bootstrap ratio values larger than the 0 . 99-quantile ( right tail ) was computed and plotted in Fig 6C ., The strongest effects are robustly expressed at δ and lower γ frequencies , directly supporting higher phase locking in controls compared to mTBI at these frequencies ., Similarly , the number of connections in the left tail defined by the 0 . 01-quantile is plotted in Fig 6D , as a function of frequency ., These connections also support the contrast in Fig 6A , but in a reverse way , representing hyper-connectivity in the mTBI which were strongest at α frequencies ., Pair-wise connections that show decreased phase synchrony in the δ and lower γ bands in mTBI are depicted in Fig 7 ., The bootstrap ratio values were averaged across wavelet frequencies within corresponding frequency bands ., A threshold of >1 was used for the figures to emphasize the spatial distribution ., Reduced δ and γ resting phase synchrony in mTBI was most pronounced between occipital areas and other brain regions , and also preferentially involved temporal lobe connections ., Similar to Figs 7 and 8A was created with a threshold of <−1 , and shows the distribution of pair-wise connections associated with increased phase synchrony in mTBI at α frequencies ., It is interesting to note that the distribution of connections which carry discriminative information between mTBI and controls , as illustrated on the transparent brain in the MNI space ( Fig 5 ) and its matrix version ( Fig 8B ) , is part of the spatial pattern representing hyper-connectivity of α rhythms in mTBI ( Fig 8A ) , which involved numerous temporal and parietal connections ., To quantitatively compare the contribution of individual frequency bands to the contrast depicted in Fig 6A , we performed a series of steps testing difference in proportions ., First , we identified the wavelets closest to the central frequencies of five canonical frequency bands: 2 Hz ( delta ) , 6 Hz ( theta ) , 11 Hz ( alpha ) , 23 Hz ( beta ) , 48 Hz ( lower gamma ) ., Then , for a given z-sore threshold ( 1% tails ) , at each central wavelet frequency , we counted connections ( out of the total 90*89/2 = 4005 ) within the positive and negative tails of the overall distribution of z-scores ( see Fig 6C and 6D ) ., For the effects defined by the tail with negative z-scores , where we observed a peak around 8 Hz ( Fig 6D ) , we ran two-sample proportion z-tests between the alpha and other frequencies ., Specifically , we tested if the numbers of connections within the negative tail at two frequency points were statistically different ., We found that the number of connections was significantly higher at alpha relative to delta ( p = 0 . 0013 ) , beta ( p = 0 . 0238 ) , and lower gamma ( p<0 . 0001 ) , but not theta ( p = 0 . 757 ) ., For the positive tail of z-scores ( Fig 6D ) , where we identified two peaks around 2 Hz and 75 Hz , we performed a series of similar two-sample proportion z-tests ., We found that the number of connections from the positive tail was statistically higher at delta relative to theta ( p = 0 . 0035 ) and alpha ( 0 . 0013 ) , whereas the number of connections at gamma was higher than theta ( p<0 . 001 ) , alpha ( p<0 . 001 ) , and beta ( 0 . 0015 ) , but not delta ( p = 0 . 1211 ) ., In addition , Fig 9 provides an example of the effects shown in Fig 6C and 6D , indicating the range of absolute values of PLV for specific connections at the characteristic frequencies ., Specifically , Fig 6 depicts a spatiotemporal interplay between synchronizations and de-synchronizations in the delta , gamma , and alpha frequency bands , and we chose three connections with the largest z-scores to illustrate the effects:, i ) between the left middle occipital gyrus ( Occipital Mid L ) and the left median cingulate and paracingulate gyri ( Cingulum Mid L ) at 2 Hz;, ii ) between the temporal pole of the left middle temporal gyrus ( Temporal Pole Mid L ) and the left gyrus rectus ( Rectus L ) at 8 Hz; and, iii ) between the left inferior temporal gyrus ( Temporal Inf R ) and the right calcarine fissure and surrounding cortex ( Calcarine R ) at 75 Hz ., Finally , we explored the effect of the length of time between injury and scan acquisition on resting MEG connectivity ., We applied the behavioural PLS analysis to correlate the phase locking value with the time between brain injury and scanning ., PLS analysis revealed a significant latent variable ( LV ) with p = 0 . 016 , which is plotted in Fig 10A as an overall correlation ( first component of LV ) and a distribution of all the bootstrap ratio values ( second component of LV ) , each associated with a unique combination of frequency and source pairing ., The right ( red ) and left ( blue ) tails of the histogram in Fig 10B represent robust positive and negative correlations , respectively , between the length of time between injury and scan and phase synchronization ., Frequency-specific number of connectins in these tails are shown in Fig 10C and 10D , respectively ., As can be seen from Fig 10D , the effect for negative correlations between the connectivity at alpha frequencies and time of scanning is strongest at alpha frequencies ., In other words , the more time that has passed since injury , the less connectivity we observed in the alpha frequency band ., It is worth noting that mTBI patients , when compared to controls , were characterized by increased connectivity at alpha frequencies ( Fig 6D ) ., The present study provides the first evidence for altered resting state neuromagnetic phase synchrony in a group of patients with mTBI , and showed that these alterations were associated with the amount of time elapsed between injury and scan acquisition ., More importantly , we demonstrate that atypical MEG network connectivity , in combination with SVM learning , can accurately detect mTBI ., This is an important step forward as mTBI is typically not detectible using conventional imaging ., Our findings indicate that neurophysiological network imaging using MEG may provide an objective method for detection of mTBI ., Moreover , we show that the distance of individual participants from the classification decision boundary was correlated with clinical symptom severity ., These results demonstrate that MEG imaging of resting state functional connectivity may offer new approaches for assessing and tracking injury severity in mTBI ., Using a data-driven approach , we showed that group differences can be characterized in terms of interplay between synchronizations and desynchronizations at different frequencies ., Specifically , we observed more increases in connectivity around theta/alpha frequencies in mTBI , whereas more decreases in connectivity in mTBI were detected for delta rhythms ., This fits the hypothesis that processing of information in the brain requires both phase synchrony and phase scattering ., Speculatively , phase synchronization can be viewed as a mechanism for long-range integration , whereas phase scattering can be a strategy to allow different local neural ensembles to share the same frequency channel by assigning specific neural signals to their own timeslots ., Furthermore , we also found that the length of time elapsed between injury and scan tended to be negatively correlated with alpha synchronization and positively correlated with delta connectivity ., These results may indicate that brain plasticity , a fundamental property for functional recovery from brain injury 20 , may potentially be described in terms of redistribution of phase synchronyzation and phase scattering at different rhythms ., A similar pattern of the interplay between increases and decreases in functional connectivity was reported in an MEG study of TBI patients in two conditions: following an injury and after a rehabilitation treatment 21 ., Noticeably , the study reported an opposite pattern , as increases in connectivity at higher frequencies such as alpha and beta , and conversely decreases in connectivity for delta and theta rhythms were associated with recovery from TBI ., One of the key differences between the two studies was the time since injury ., In our study , MEG data were recorded from mTBI patients , who were all within 3 months of injury ( on average , one month ) ., In 21 , the mean time since injury was almost 4 months , and the rehabilitation program lasted for about 9 months ., Prior studies have indicated that resting state MEG can be used to detect mild and moderate TBI at the level of single individuals , but rather than focusing on inter-regional oscillatory synchrony , such research focused on the regional expression of excessive slow-wave activity 6 , 7 ., It has been proposed that axonal sheering caused by rapid deceleration and rotational forces plays a critical role in the pathology of TBI as well as its impact on functional networks and cognition 14 ., Interestingly , regional expression of increased slow-wave activity has been shown to be either proximal to white matter abnormalities revealed by DTI , or in some cases , remote if micro-structural abnormalities occur in a major tract innervating that region 8 ., Furthermore , this implies that excessive slow-wave activity reported in prior studies may be related to alterations in functional connectivity reported in the present investigation ., Recent evidence indicates that regional concentrations of oscillatory slowing also correspond to particular symptoms expressed 7 , raising the question of whether region-specific differences in functional connectivity may relate to specific patterns in post-concussive symptoms ., Research using EEG has also reported that electrophysiological interactions among brain regions are atypical in mTBI ., Reduced inter-hemispheric phase synchrony among EEG scalp electrodes has been reported , and it was shown that such connectivity reductions in the beta and gamma frequency ranges were associated with alterations in white matter microstructure 10 ., The network organization of resting state EEG connectivity has also been shown to be altered in mTBI 22 ., An MEG investigation of patients with mild , moderate and severe TBI reported functional network disconnection in this group 23 ., Using the data set employed in the present study , we previously showed that resting state correlations in the amplitude envelope of MEG activity is elevated in the delta , theta and alpha bands in mTBI , and that these alterations are associated with cognitive and affective sequelae in this group 24 ., Interestingly , this pattern of alteration is different from MEG network alterations associated with PTSD ( which is often a co-morbidity of mTBI ) which was associated with high-frequency increases in resting phase synchrony 25 ., Neural oscillations and their synchronization among brain areas are thought to play a critical role in cognition 11 , 21 , and resting neuromagnetic synchrony and amplitude correlations are presently thought to reflect intrinsic functional networks underpinning cognition , perception and their disturbance in clinical populations 13 ., EEG research has also indicated that reduced electrophysiological interactions among brain areas may contribute to cognitive and behavioural problems associated with PCS ., Reduced EEG coherence , for example , has been observed during visuospatial working memory in mTBI 26 and disrupted organization of network synchronization during episodic memory processing has also been reported 27 ., Such reports of altered task dependent connectivity are congruent with reports of atypical electrophysiological and hemodynamic responses during cognitive processing following mTBI 28 ., MRI studies have indicated altered functional network connectivity in mTBI 29–31 , in the very low hemodynamic frequency oscillations measured by fMRI , which have been related to cognitive problems and recovery in this group 32 ., During resting state , fMRI abnormalities have been reported which encompass visual , limbic motor and cognitive networks 29 ., Altered default mode network connectivity 32 and regulation have been reported in mTBI ., Spontaneous BOLD correlations have also been shown to be atypical in thalamocortical networks in mTBI patients , and these alterations are correlated with both clinical symptomatology and cognitive performance 30 ., That altered connectivity is prominent in both neurophysiological and hemodynamic imaging studies is not surprising , as damage to white matter tracts in the form of diffuse axonal injury is common in severe brain injury 32–35 ., Investigations of brain microstructure in such populations indicate altered axonal structure in both gray and white matter 36–38 ., The present study capitalizes on rapidly emerging methods combining analysis of brain network connectivity with machine learning approaches supporting classification at the level of individual participants ., This provides new insights into complex spatiotemporal shifts in intrinsic coupling in neurophysiological brain networks following mTBI ., More importantly , the present work provides potentially clinically translatable methods that will permit the detection of mTBI in single individuals where conventional radiological imaging approaches are inconclusive ., The finding that classification confidence is associated with self-reported symptom severity indicates that these methods may provide quantitative and objective measurements of brain changes underlying PCS ., This could have significant impact on current clinical practice ., An objective , quantitative method for diagnosing brain dysfunction after mTBI would allow identification of patients at risk for a subsequent injury , be invaluable for developing parameters around return to play / work / duty , and assist in developing guidelines for providing care , monitoring treatment efficacy and tracking recovery ., MEG data were recorded from 20 men with mTBI ( 21–44 years of age , mean = 31±7 years , 2 left-handed ) , all within three months of injury ( days since injury = 32 ± 18 days ) ., Participants with mTBI were recruited through the Emergency Department of Sunnybrook Health Science Centre in Toronto ., The inclusion criteria were: concussion symptoms while in emergency; Glasgow Coma Scale ≥13 ( within 24 hours of injury ) ; if loss of consciousness occurred , then less than 30min; if post-traumatic amnesia occurred , then less than 24 hours; causes of head injury were clear ( e . g . , sustaining a force to the head ) ; no skull fracture; no abnormalities on Computer Tomography ( CT ) scan and no previous incidences of concussion ., Participants in the mTBI group completed the Sports Concussion Assessment Tool 2 ( SCAT2 ) Symptom Checklist and Symptom Severity Score; were able to tolerate the enclosed space of the MRI; were English speaking and able to complete tasks during MEG and MR scans and able to give informed consent ., The mean Severity score of mTBI patients was 20 ± 19 , whereas the Symptom score was 9 ± 6 ., The MEG and MRI scans were obtained , on average , on 32nd day since injury: 32 ± 18 days ., Potential participants were screened prior to recruitment and none of the mTBI participants reported any post-traumatic stress disorder , neurological or psychiatric symptoms , and psychoactive medication use ., All of the MRI scans were read by a neuroradiologist , and there were no abnormalities noted ., An age- and sex-matched control group without any history of TBI included 21 participants ( 20–39 years of age , mean = 27±5 years , 1 left-handed ) ., The control group had no history of TBI ( mild , moderate or severe ) , no neurological or psychiatric disorders , and were not on psychoactive medications ., None of the participants had MRI contraindications such as metallic implants or metal dental work ., Data acquisition was performed with the informed consent of each individual and with the approval of the Research Ethics Board at the Hospital for Sick Children ( SickKids ) ., MEG data were acquired in a magnetically shielded room at SickKids using a whole-head CTF system ( MISL Ltd . , Coquitlam , BC , Canada ) with 151 axial gradiometers as well as reference sensors for gradient correction ., For each subject , 5 minutes of MEG data were continuously recorded at 600Hz using third-order spatial gradient noise cancellation ., 60Hz and 120Hz notch filters were applied to MEG recordings ., Data were also band-pass filtered between 1Hz and 150 Hz with a fourth-order Butterworth digital filter applied first in a forward , and then in a reverse direction so as to produce zero phase distortion ., Head position during testing was monitored via three localization coils , positioned at the nasion , and the left and right pre-auricular points ., Anatomical MRI was performed on the same day at SickKids on a 3T MR scanner ( MAGNETOM Tim Trio , Siemens AG , Erlangen , Germany ) with a 12-channel head coil ., The three fiducial coils used in the MEG were replaced with radio-opaque markers for all participants ., These markers can be seen on their T1-weighted images for co-registration of the MEG source locations to the MRI images ., Anatomical images were collected by whole-brain T1-weighted MRI scans ( 3D SAG MPRAGE: GRAPPA = 2 , TR/TE/TI/FA = 2300/2 . 96/900/9 , FOV/Res = 192x240x256 , 1mm isotropic voxels ) ., Individual MRI scans were normalized into Montreal Neurological Institute ( MNI ) space based on the ICBM 2009c Nonlinear Symmetric 1 × 1 × 1mm template 39 ., We applied a nonlinear diffeomorphic registration , as implemented in the ANTS toolbox 40 , 41 ., This transformation to MNI space was additionally used to warp a manually segmented inner skull surface from the MNI ICBM template to subject space ., Using this inner skull surface , a multi-sphere head model was fit for each subject 42 ., MEG data were co-registered to each participant’s individual anatomical MRI to constrain neuromagnetic sources to subject-specific head shape and structural anatomy ., To reconstruct neuromagnetic source activity , we first selected 90 seed locations in MNI space , which represented all cortical and subcortical brain regions in the Automated Anatomical Labeling ( AAL ) atlas 43 ., Regions specified by the AAL atlas and located in the cerebellum were excluded from the further analysis ., For visualization purposes , the regions were re-ordered according to which lobe each region belongs to ., The new order of the regions is given in Table 1 ( the left region goes first , followed by the right one ) ., Specifically , for each region from the AAL parcellation , the seed location was defined as a voxel within the region , which was closest , in the mean-square sense , to the means of x- , y- , and z-coordinates , averaged across all the voxels in this brain region 44 ., Source estimation was performed at these 90 locations , using an adaptive spatial filter ( vector beamformer ) 45 ., For each subject , 27 non-overlapping epochs of 10 seconds duration were extracted such that head motion within each epoch did not exceed 3mm in any direction for any of three head location coils ., The time-frequency representation of the original time series for each reconstructed source was derived from the wavelet decomposition , using a time-frequency toolbox 46 ., Thirty frequency points equally spaced on a logarithmic scale were selected to cover the range between 1Hz and 75Hz ., The analysis of phase synchronization between the neuromagnetic sources was performed on spectrally decomposed data ., We computed phase-locking values 47 , which are known in the literature under different names such as mean phase coherence 48 or phase synchronization index 49 ., Phase synchronization emerged from studying coupled nonlinear systems 50 , and is based on an idea that the existence of correlations between the phases of coupled systems does not imply correlation between their amplitudes ., A common method for obtaining phase dynamics for analyzing phase synchronization between brain signals is based on wavelet transformation 51 ., A signal can be decomposed into a set of brief oscillatory patterns called wavelets ., Specifically , wavelet coefficients Wx ( τ , f ) at time τ and frequency f are obtained by convolving a given signal x ( t ) with a zero-mean special function or wavelet ψτ , f ( t ) :, Wx ( τ , f ) =∫−∞+∞x ( t ) ψτ , f ( t ) dt, ( 1 ), where ψτ , f ( t ) is a short segment of a oscillatory signal ( wavelet ) obtained from an elementary function called the mother wavelet by dilutions and translations ., Often , a specific form of the mother wavelet is used , known as the Richer wavelet or Mexican hat function , which is defined as the negative normalized second derivative of a Gaussian function ., To decompose a signal at a specific frequency f and time τ , the mother wavelet is compressed or dilated , and then translated such that ψτ , f ( t ) is centered at time τ ., To maintain a consistent frequency resolution , the bandwidth of the envelope is set to be inversely proportional to f , such that each wavelet contains the same number of cycles ., In general , the coefficients Wx ( τ , f ) are complex numbers ., The transformation Eq ( 1 ) thus defines both the amplitude of signal x ( t ) and the phase over a range of times τ and frequencies f ., The instantaneous phase ϕx ( τ , f ) is the angular component ( phase angle ) of Wx ( τ , f ) ., The relative phase Δϕx ( τ , f ) of two signals , x ( t ) and y ( t ) , is defined as a time series of the difference between the instantaneous phase of each signal , namely, Δϕx , y ( τ , f ) =ϕx ( τ , f ) −ϕy ( τ , f ), ( 2 ), which can be computed from the wavelet coefficients at time τ and frequency f from, eiΔϕx , y ( τ , f ) =Wx ( τ , f ) Wy* ( τ , f ) |Wx ( τ , f ) ||Wy ( τ , f ) |, ( 3 ), where Wy* ( τ , f ) is the complex conjugate of Wy ( τ , f ) ., The phase differences can be projected as a series of two-dimensional vectors onto the unit circle , one per time point τ = τ1 , … , τN ., The phase-locking value PLVx , y ( f ) , which reflects the amount of phase-synchrony between two signals across time , is computed as the length of the resultant ( mean ) vector:, PLVx , y ( f ) =〈eiΔϕx , y ( τ , f ) 〉τ=|1N∑k=1NeiΔϕx , y ( τk , f ) |, ( 4 ), By construction , PLVx , y ( f ) is limited between 0 and 1 ., When the relative phase distribution is concentrated around the mean , the PLV is close to one , whereas phase scattering will result in a random distribution of phases and PLV close to zero ., For each epoch , for all pairs of 90 regions of interest ( ROIs ) , frequency-specific phase differences were computed as functions of time ., The phase-locking value , PLVx , y ( f ) , was calculated as relative stability of the phase differences between two signals at a given frequency , subsequently averaging across epochs ., Thus , 30 90-by-90 matrices were produced for each subject , representing functional connectivity in terms of phase-locking between 90 neuromagnetic sources at 30 frequency points ., In the present study , Support Vector Machine ( SVM ) learning was used to predict the clinical status Y of a subject ( mTBI or control ) from a set of features X obtained from the subject’s MEG data 52 ., These features are represented by | Introduction, Results, Discussion, Materials and Methods | Accurate means to detect mild traumatic brain injury ( mTBI ) using objective and quantitative measures remain elusive ., Conventional imaging typically detects no abnormalities despite post-concussive symptoms ., In the present study , we recorded resting state magnetoencephalograms ( MEG ) from adults with mTBI and controls ., Atlas-guided reconstruction of resting state activity was performed for 90 cortical and subcortical regions , and calculation of inter-regional oscillatory phase synchrony at various frequencies was performed ., We demonstrate that mTBI is associated with reduced network connectivity in the delta and gamma frequency range ( >30 Hz ) , together with increased connectivity in the slower alpha band ( 8–12 Hz ) ., A similar temporal pattern was associated with correlations between network connectivity and the length of time between the injury and the MEG scan ., Using such resting state MEG network synchrony we were able to detect mTBI with 88% accuracy ., Classification confidence was also correlated with clinical symptom severity scores ., These results provide the first evidence that imaging of MEG network connectivity , in combination with machine learning , has the potential to accurately detect and determine the severity of mTBI . | Detecting concussion is typically not possible using currently clinically used brain imaging , such as MRI and CT scans ., Magnetoencephalographic ( MEG ) imaging is able to directly measure brain activity at fast time scales , and this can be used to map how various areas of the brain interact ., We recorded MEG from individuals who had suffered a concussion , as well as control subjects who had not ., We found characteristic alterations of inter-regional interactions associated with concussion ., Moreover , using a machine learning approach , we were able to detect concussion with 88% accuracy from MEG connectivity , and confidence of classification correlated with symptom severity ., This potentially provides new quantitative and objective methods for detecting and assessing the severity of concussion using neuroimaging . | traumatic injury, medicine and health sciences, diagnostic radiology, neural networks, brain damage, brain electrophysiology, electrophysiology, neuroscience, magnetic resonance imaging, artificial intelligence, brain mapping, bioassays and physiological analysis, neuroimaging, electroencephalography, research and analysis methods, computer and information sciences, imaging techniques, clinical neurophysiology, traumatic brain injury, electrophysiological techniques, critical care and emergency medicine, trauma medicine, radiology and imaging, diagnostic medicine, neurology, physiology, biology and life sciences, magnetoencephalography, cognitive science, neurophysiology, machine learning | null |
journal.pgen.1003455 | 2,013 | Improved Detection of Common Variants Associated with Schizophrenia and Bipolar Disorder Using Pleiotropy-Informed Conditional False Discovery Rate | Converging evidence suggests that complex human phenotypes are influenced by numerous genes each explaining a small proportion of the variance 1 ., Though thousands of single nucleotide polymorphisms ( SNPs ) have been identified by genome-wide association studies ( GWAS ) 2 , 3 , these SNPs fail to explain a large proportion of the heritability of most complex phenotypes studied ., This is commonly referred to as the ‘missing heritability’ problem ., Recent findings indicate that GWAS have the potential to explain a greater proportion of the heritability of common complex phenotypes 4–6 , and more SNPs are likely to be identified in larger samples 7 ., Due to the polygenic nature of most complex traits and disorders , a large number of SNPs are likely to have associations too small in magnitude to be identified with currently available sample sizes 8 ., New analytical methods are therefore needed to reliably identify a larger proportion of SNPs associated with complex diseases and phenotypes , since recruitment and genotyping of sufficiently large samples for existing methods may be impractical and prohibitively expensive ., Genetic pleiotropy is defined as a single gene or variant being associated with more than one distinct phenotype ., In the present study we use a new genetic pleiotropy-informed approach for GWAS to capture more of the polygenic effects in complex phenotypes ., Given the high number of traits in humans , and the relatively small number of genes ( ∼20 , 000 ) , some genes have to affect multiple traits ( genetic pleiotropy ) 10 ., By combining independent GWAS from associated disorders , we hypothesize that for disorders with related etiologies a genetic pleiotropy-informed approach can significantly improve gene discovery and help capture more of the missing heritability ., Recent findings suggest overlapping SNPs between several human traits 9 , 11 and disorders 12–14 ., To date , methods to assess this genetic pleiotropy have not taken full advantage of the existing GWAS data and the majority of studies have focused on the subset of SNPs exceeding a Bonferroni-corrected threshold of significance for each trait or disorder 12–14 ., However , this approach cannot detect SNPs that only reach genome-wide significance in the combined analysis but do not meet Bonferroni-corrected significance in the individual phenotype ( hereafter referred to as polygenic pleiotropy ) ., Combining GWAS statistics from two disorders also provides increased power to discover genes associated with common biological mechanisms , and thus inform on overlapping pathophysiological relationships between the disorders ., In the current study , we use a pleiotropy-informed statistical approach to improve gene discovery in schizophrenia and bipolar disorder , two disorders with high heritability 15 , where most of the underlying genetic architecture remains unknown 13 , 14 , despite recent discoveries 13 , 14 , 16 , 17 ., Schizophrenia and bipolar disorder share several clinical characteristics 18–20 , including psychotic symptoms , disorders of thought and impairment of cognitive functions 21 ., The disorders are often also treated with similar pharmacological agents 18 , 19 ., Whether schizophrenia and bipolar disorder should be regarded as separable disease entities or as a single disease with a spectrum of symptoms 18–20 , as proposed in the continuum hypothesis of psychosis 22 , has been much discussed ., With the forthcoming revision of the Diagnostic and Statistical Manual of Mental Disorders ( DSM ) , this question has received renewed attention 19 , 20 , 23 ., Both disorders have an estimated heritability of 0 . 7–0 . 8 , and are regarded as complex disorders with a polygenic architecture ., Several lines of evidence have suggested overlapping genetic susceptibility in bipolar disorder and schizophrenia 15 , 24–26 ., Recently , a combined analysis of two large GWAS ( 16 , 374 cases and 12 , 044 controls ) revealed three loci ( CACNA1C , rs4765905 , p\u200a=\u200a7 . 0×10−9 , ANK3 rs10994359 , p\u200a=\u200a2 . 5×10−8 , ITIH3/4 region rs2239547 , p\u200a=\u200a7 . 8×10−9 ) significantly associated with both disorders ( Fishers combined p in combined samples ) 13 , 14 ., Still , given the high degree of heritability and large similarities in clinical phenotypes , there are likely several more undiscovered overlapping genetic factors ., Here , using summary statistics from two independent large GWAS , we applied a model-free statistical analysis method to identify SNPs exhibiting pleiotropic relationships between schizophrenia and bipolar disorder ., First , we separated out the common controls in the bipolar disorder and schizophrenia samples 13 , 14 , ensuring non-overlapping samples ., After applying genomic inflation control , we computed the conditional empirical cumulative distribution functions ( cdfs ) of the corrected p-values ., Empirical cdfs for schizophrenia SNP p-values were determined conditional on the significance of the corresponding nominal p-values in bipolar disorder , and vice versa ., For each nominal p-value , an estimate of the conditional False Discovery Rate ( FDR ) was obtained from the conditional empirical cdfs 5 ., Using this conditional FDR method , we constructed two-dimensional FDR “look-up” tables , with FDR in schizophrenia SNPs computed conditional on nominal bipolar disorder p-values , and vice versa ., Using these tables we identified 58 loci associated with schizophrenia and 35 loci associated with bipolar disorder at a conditional FDR level of 0 . 05 ., We used a conjunction method to investigate SNPs significantly associated with both schizophrenia and bipolar disorder ., Specifically , we computed the conditional FDR for schizophrenia given bipolar disorder nominal p-values , and conditional FDR for bipolar disorder given schizophrenia nominal p-values , and took the maximum of both values as the conjunction FDR ., With this approach we identified 14 pleiotropic loci indicating several overlapping genetic risk factors for the two disorders ., Finally , using mixture model-based analyses we estimated the proportion and distribution of non-null SNPs , demonstrating that the large increase in power from using conditional vs . unconditional FDR methods is derived from the high polygenicity of both phenotypes with many test statistics just below significance thresholds , and the largely overlapping distribution ( high degree of pleiotropy ) of non-null SNPs for schizophrenia and bipolar disorder ., Under large-scale testing paradigms , such as GWAS , quantitative estimates of likely true associations can be estimated from distributions of summary statistics 27 , 28 ., A common method for visualizing the ‘enrichment’ of statistical association relative to that expected under the global null hypothesis is through Q-Q plots of nominal p-values obtained from GWAS summary statistics ., The usual Q-Q curve has the nominal p-value , denoted by “p” , as the y-ordinate and the corresponding value of the empirical cdf , here denoted by “q , ” as the x-ordinate ., Under the global null hypothesis the theoretical distribution is uniform on the interval 0 , 1 ., As is common in GWAS , we instead plot −log10 p against −log10 q to emphasize tail probabilities of the theoretical and empirical distributions ., As such , genetic ‘enrichment’ refers to a leftward shift in the Q-Q curve , corresponding to a larger fraction of SNPs with nominal −log10 p-value greater than or equal to a given threshold ., Conditional Q-Q plots are formed by creating subsets of SNPs based on values of an additional variable ( auxiliary measure ) for each SNP , and computing Q-Q plots separately for each subset of SNPs ., If SNP enrichment is captured by variation in the auxiliary measure , this is expressed as successive leftward deflections in conditional Q-Q plots as values of the additional variable increase ., Conditional Q-Q plots for schizophrenia given nominal p-values of association with bipolar disorder ( SCZ|BD; Figure 1A ) show enrichment across different levels of significance for bipolar disorder ., The earlier departure from the null line ( leftward shift ) suggests a greater proportion of true associations for a given nominal schizophrenia p-value ., Successive leftward shifts for decreasing nominal bipolar disorder p-value thresholds indicate that the proportion of non-null effects in schizophrenia varies considerably across different levels of association with bipolar disorder ., For example , the proportion of SNPs in the −log10 ( pBD ) ≥3 category reaching a given significance level for schizophrenia ( e . g . , −log10 ( pSCZ ) ≥4 ) is roughly 50 times greater than for the −log10 ( pBD ) ≥0 category ( all SNPs ) , indicating a high level of enrichment ., An even stronger pleiotropic enrichment can be seen for bipolar disorder conditioned on nominal p-values of association with schizophrenia ( BD|SCZ; Figure 1B ) , Here , the proportion of SNPs in the −log10 ( pSCZ ) ≥3 category reaching a given significance level for bipolar disorder ( e . g . , −log10 ( pBD ) ≥4 ) is roughly 500 times greater than for the −log10 ( pSCZ ) ≥0 category ( all SNPs ) , indicating a very high level of enrichment ., We constructed a “conditional” Manhattan plot for schizophrenia showing the FDR conditional on bipolar disorder ( Figure, 2 ) and identified significant loci on a total of 18 chromosomes ( 1–4 , 6–16 , 18 , 20 and, 22 ) associated with schizophrenia leveraging the reduced FDR obtained by the associated bipolar disorder phenotype ., To estimate the number of independent loci , we ‘pruned’ the associated SNPs ( removed SNPs with linkage disequilibrium ( LD ) >0 . 2 ) , and identified a total of 58 independent loci with a significance threshold of conditional FDR<0 . 05 ( Table 1 ) ., Using the more conservative conditional FDR threshold of 0 . 01 , 9 independent loci remained significant ., One locus was located in the HLA region on chromosome 6 ., Of note , using a standard Bonferroni-corrected approach , no loci would have been discovered ., Using the FDR method in schizophrenia alone , 4 loci were identified ., Of these , the regions close to TRIM26 ( 6p21 . 3 ) , MMP16 ( 8q21 . 3 ) and NT5C2 ( 10q24 . 32 ) have been identified in earlier GWAS studies after including large replication samples 13 ., The remaining loci would not have been identified in the current sample without using the pleiotropy-informed conditional FDR method ., Of interest , the VRK2 region ( 2p16 . 1 ) was identified in the previous sample after including a large schizophrenia replication sample 30 , and the ITIH4 region ( 3p21 . 1 ) , ANK3 ( 10q21 ) and CACNA1C ( 12p13 . 3 ) were discovered previously in the same , combined schizophrenia and bipolar disorder sample 13 , 14 ., Thus , the current pleiotropy-informed FDR method validated 7 loci discovered in considerably larger samples , and discovered 51 new loci ., We constructed a “conditional” Manhattan plot for bipolar disorder showing the FDR conditional on schizophrenia ( Figure, 3 ) and identified significant loci on a total of 16 chromosomes ( 1–3 , 5–8 , 10–14 , 16 and 19–22 ) associated with bipolar disorder leveraging the reduced FDR obtained by the associated schizophrenia phenotype ., To estimate the number of independent loci , we pruned the associated SNPs ( removed SNP with LD >0 . 2 ) , and identified a total of 35 independent loci with a significance threshold of conditional FDR<0 . 05 ( Table 2 ) ., Of these , one locus was complex , i . e . included several significant SNPs , and the rest were single gene loci ., Using the more conservative conditional FDR threshold of 0 . 01 , 5 independent loci remained significant ., The most significant locus was close to ANK3 on chromosome ( 10q21 ) ., This is the only locus that would have been discovered using standard methods based on p-values ( Bonferroni correction ) ., Using the FDR method in bipolar disorder alone , an additional locus was identified , close to CACNA1C ( 12p13 . 3 ) ., Both these loci have been discovered earlier 14 , 31 ., The remaining 33 loci would not have been identified in the current sample without using the pleiotropy-informed conditional FDR method ., Of these , the regions close to SYNE1 ( 6q25 ) and ODZ4 ( 11q14 . 1 ) have been identified in earlier GWAS after including large replication samples 14 , 32 ., Of interest , the ITIH3 region ( 3p21 . 1 ) , ANK3 ( 10q21 ) and CACNA1C ( 12p13 . 3 ) were discovered previously in the same , combined schizophrenia and bipolar disorder sample 13 , 14 ., Thus , pleiotropy-informed conditional FDR validated 5 loci discovered in considerably larger samples , and discovered 30 new loci ., To identify pleiotropic loci in schizophrenia and bipolar disorder , we performed a conjunction FDR analysis , using this to construct a “conjunction” Manhattan plot ( Figure 4 ) ., We detected 14 independent pleiotropic loci ( pruned based on LD>0 . 2 , black line around large circles ) with conjunction FDR<0 . 05 , all single gene loci , located on a total of 10 chromosomes ( chr . 1 , 3 , 6 , 7 , 10 , 12 , 14 , 16 , 20 , 22 – for further details , please see Table 3 ) ., Of these loci , 3 have been implicated in bipolar disorder and schizophrenia earlier: NOTCH4 ( 6p21 . 2 ) with schizophrenia using a larger replication sample 13 , 17 , and the ITIH4 ( 3p21 . 1 ) , and CACNA1C ( 12p13 . 3 ) regions , both discovered previously in the same , combined schizophrenia and bipolar disorder sample 13 , 14 ., Interestingly only one conjunctional locus was found on chromosome 6 , suggesting that there are several schizophrenia loci on this chromosome not overlapping with bipolar disorder ., The ANK3 locus was not indicated in the conjunctional analysis , which indicates that the overlap is mostly driven by the association in bipolar disorder ( Table 2 ) ., The direction of the effect ( z-scores ) across all the pleiotropic SNPs was the same for bipolar disorder and schizophrenia , except for locus 33 ( BC039673 , 20p13 ) , which could be due to differences in LD structure in this region ., These findings suggest overlapping genetic pathways in schizophrenia and bipolar disorders ., Our model-free conditional FDR analyses circumvent the issue of bias due to model misspecification ., However , to ascertain the impact of effective sample size and conditioning on relative power over using unconditioned FDR on current sample sizes , it is necessary to use a model-based approach that estimates the proportion and distribution of non-null SNPs 33 ., We thus posit a mixture of null and non-null Gaussian distributions 34 ( see Methods and Text S1 ) ., Resulting model fits are displayed in Figure 5 for schizophrenia and bipolar disorder for absolute z scores ≥3 ., Left panels are actual data , whereas right panels are hypothetical realizations from a doubling of effective sample size , generated from mixture model fits ., Null densities largely coincide with the overall densities except for z scores with absolute value larger than 4 , at which point the ratio of null to total SNPs , equal to the local false discovery rate ( local FDR ) , is less than 0 . 5 ( left panels of Figure 5 ) ., Thus , while highly polygenic , most non-null SNPs have local FDR much larger than 0 . 05 ., The local FDR does not drop below 0 . 05 until absolute z scores exceed 5 ., Far more of the “hidden” non-null SNPs lie below this significance threshold than above it ., Many of these hidden SNPs lie just below the significance threshold , so that an effective doubling of the sample size produces a ∼30 times increase in number of rejected non-null SNPs with local FDR ≤0 . 05 ( right panels of Figure 5 ) ., Another model-based analysis using a bivariate mixture of Gaussians showed that a very high proportion of the non-null schizophrenia SNPs are also non-null for bipolar disorder ( and vice versa ) leading to large increases in power when using the conditional FDR approach ., This increase in power is also due to the large number of SNPs with p-values just below the Bonferroni threshold ., Figure 6 shows the power , or sensitivity to detect non-null SNPs for differing local FDR cut points from unconditional and conditional local FDR and , for comparison , from a hypothetical doubling of the number of subjects ., Using conditional over unconditional local FDR results in an increase of 15–20 times the number of non-null SNPs discovered for a local FDR≤0 . 05 ., The increase in power for conditional FDR , while dramatic , is not as large as what would be obtained by doubling the sample size ., This is not unexpected , given that the highly polygenic non-null SNPs for schizophrenia and bipolar disorder , many just below the given significance thresholds , are largely but not completely overlapping ., Note , given their highly polygenic distribution the vast majority of non-null SNPs remain undiscovered even using conditional FDR approaches or under an effective doubling of the number of subjects ., To test for enrichment with a “control trait” with little or no polygenic overlap with psychiatric disease , we performed pleiotropy analysis using type 2 diabetes ( T2D ) GWAS data ., The analyses confirmed that there was a very small level of pleiotropic enrichment between schizophrenia and T2D , leading to little if any improvement in statistical power ( See Text S1 and Figure S7 ) ., In the present study we leveraged the power of GWAS data from two independent schizophrenia and bipolar disorder samples , and demonstrate how GWAS from associated psychiatric disorders can improve discovery of novel susceptibility loci ., Using standard GWAS analytical methods , we identified only one significant locus ., By applying traditional FDR methods in the separate GWAS samples , we found an additional 6 loci ( 2 in bipolar disorder , 4 in schizophrenia ) ., Combining the independent schizophrenia and bipolar disorder GWAS samples , we identified a total of 58 loci in schizophrenia and 35 in bipolar disorders , with conditional FDR<0 . 05 as a threshold ., Nine of the current loci have been identified earlier in larger samples using standard GWAS analytical methods ( 7 in schizophrenia , 5 in bipolar disorder , and 3 in combined samples ) , while 10 other loci have been reported to show borderline association with bipolar disorder or schizophrenia ( Table S1 ) ., These results demonstrate the feasibility of using a cost-effective , pleiotropy-informed conditional FDR approach to discover common variants in schizophrenia and bipolar disorders ., The proposed statistical approach is based on the observation that all SNPs should not be treated as exchangeable ., Rather , a SNP with large effects in two associated phenotypes has a higher probability of being a true non-null effect , and hence also a higher probability of being replicated in independent studies ., We thus applied a conditional FDR approach we have previously developed for GWAS p-values 36 , adapted from methods originally used for linkage analysis and microarray expression data 5 , 37 ., Decreased conditional FDR ( equivalently , increased conditional TDR ) for a given nominal p-value increases power to detect true non-null effects ., Increased conditional TDR is directly related to increased replication effect sizes and replication rates in de novo samples ., Using this conditional approach we were able to increase power to detect true non-null signals in independent studies for given nominal p-values cut-offs ., Equivalently , in the conditional approach the FDR can be used to control FDR at a given level while increasing power to discover non-null SNPs over approaches that treat all SNPs as interchangeable ., We also applied a previously developed conjunction FDR approach 36 to investigate which SNPs are pleiotropic , impacting risk of both schizophrenia and bipolar disorder ., The conjunction statistic used is the maximum of the conditional FDR for schizophrenia given bipolar disorder and vice versa ., SNPs that exceed a stringent conjunction threshold are thus highly likely to be non-null in the two phenotypes simultaneously ., The extra number of significant loci identified in the current study compared to ‘conventional’ GWAS methods is remarkable ., The power analyses suggest that the large increase in power is due to the conditional FDR method , and not an implicit higher false discovery rate ., Compared to conventional GWAS methods , traditional FDR methods only identified a few extra loci ., The large increase in power came from using conditional FDR , which identified 14 . 5 times as many schizophrenia SNPs and 17 . 5 times as many bipolar SNPs ( at FDR≤ . 05 level ) compared to traditional FDR methods ., This large increase in power seems to be due to two factors: the highly polygenic distribution of non-null SNPs and the high degree of pleiotropy between schizophrenia and bipolar disorder ., We quantified this using a model-based mixture of null and non-null Gaussian distributions 34 ., Mixture models estimate roughly 1 . 2% of tag SNPs are non-null in both bipolar disorder and schizophrenia ., With over 1 million assayed SNPs in common between both phenotypes , the number of un-pruned , non-null SNPs is thus in excess of 12 , 000 in each phenotype ., The vast majority of these non-null SNPs are hidden within the large proportion ( ∼99% ) of null SNPs ., Results are in line with recent findings of a high proportion of variation in schizophrenia susceptibility captured by common SNPs 6 ., Taken together , these findings strongly suggest that Empirical Bayes methods , as outlined by Efron 27 should be the method of choice for analyzing GWAS of polygenic human phenotypes , and for leveraging pleiotropy with other complex humans traits ., The current findings of polygenic enrichment suggest that genetic pleiotropy is important in severe mental disorders , as has been indicated earlier 13–15 , 24 , 25 ., However , by using conditional FDR , we were able to leverage the overlapping polygenetic architecture to identify more of the specific SNPs involved ., The current approach identified 58 loci in schizophrenia compared to 7 in the original publication 13 ., In bipolar disorder , the added power from schizophrenia GWAS identified 35 loci compared to two loci in the original study 14 ., It is important to note that this improvement in gene discovery was obtained despite the much smaller number of controls in the current analyses because the original analyses of the two disorders used largely overlapping control samples ., Since we used data from the 1000 Genomes Project ( 1KGP ) to calculate LD structure , the number of loci can vary somewhat compared to the original analysis ., For both disorders , most of the current findings were borderline significant in the original GWAS mega-analysis , or identified in other GWAS of partly overlapping samples , such as TRANK1 38 and SYNE1 32 ., Several of the currently identified genes have been associated in previous candidate gene studies , such as DAOA 39 ., Further , we identified 14 loci strongly associated with both disorders , compared to three in the original combined analysis 13 , 14 ., Previous studies have mainly used Fisher combined tests for joint analysis , which test the null-hypothesis of no association in any phenotype , which means that the signal can be driven by one of the phenotypes ., In contrast , conjunction FDR analyses assess the evidence that either phenotype is non-null ., It is therefore difficult to directly compare the current findings with previous results ., However , of the three identified loci in previous combined analysis 13 , 14 , both the ITIH3-4 and CACNA1C regions were confirmed with the conjunctional analyses , but not the ANK3 region ., We found the latter to be associated with bipolar disorder in the current analysis , which suggests that previous results found with Fisher combined statistics were driven by the stronger association in bipolar disorder 13 , 14 ., The current findings suggest some interesting gene candidates related to overlapping biology of bipolar disorder and schizophrenia ., The Major Histocompatibility Complex loci associations with schizophrenia in previous studies 13 , 17 seem not to be strengthened by the combined analysis with bipolar disorder , as they are minimally represented among the current pleiotropic loci ( conjunction FDR analyses ) ., The only pleiotropic gene on chromosome 6 was NOTCH 4 , which has recently also been implicated in bipolar disorder 26 , ., The current findings strengthen the involvement of genes related to calcium homeostasis and receptor functioning ., In schizophrenia , both CACNA1C and ANK3 were identified , and in bipolar disorder TRANK1 and CACNB2 were also significantly associated ., CACNA1C and CACNB2 are related to key proteins involved in unifying the generation of calcium spikes in neocortical pyramidal neurons , which is a closely integrated process 41 ., It is likely that such functional processes could be involved in generation of symptoms in severe mental disorders , and may thus be a potential therapeutic target ., Interestingly , PPM1F , a Mg2+/Mn2+ dependent protein phosphatase , related to calcium/calmodulin-dependent protein kinase II gamma , was also associated with both disorders , and seems to further strengthen the hypothesis that alterations in electrophysiological function play a role in the pathophysiology of these disorders ., It is also noteworthy that SNPs located close to MAD1L1 were significantly associated with both schizophrenia and bipolar disorder ., MAD1L1 is located in a human accelerated region in the genome , which shows a large difference between humans and chimpanzees 42 , and thus is suggested to be involved in human-specific traits ., In addition to uncovering more of the missing heritability of bipolar disorder and schizophrenia , the current findings support the notion that genetic pleiotropy is important for variation in human phenotypes 9 , and suggest that there is substantial polygenic pleiotropy between bipolar disorder and schizophrenia which warrants further exploration ., In the current study we defined pleiotropy as a single gene or variant being associated with more than one distinct phenotype ( diseases ) 9 ., It is possible that some of the loci identified in the current study are not pleiotropic but rather underlie common aspects of the schizophrenia and bipolar disorder phenotypes 9 ., This possibility warrants further investigation , but requires samples with more detailed information on clinical characteristics ., In the current analyses we focused on SNPs , but gene-based pleiotropy is also of interest 10 , as is the use of the current approach for developing methods for risk prediction across traits ., However , these applications require raw data from individual participants and these data are not currently available ., In conclusion , the current findings demonstrate that in schizophrenia and bipolar disorder , pleiotropy-informed conditional FDR can improve the statistical power for detecting novel polygenic effects ., Results from conditional and conjunction FDR analyses also offer insights into potential shared mechanistic relationships between these two mental disorders ., The relevant institutional review boards or ethics committees approved the research protocol of the individual GWAS used in the current analysis and all human participants gave written informed consent ., We obtained complete GWAS results in the form of summary statistics p-values from the Psychiatric GWAS Consortium ( PGC ) – Schizophrenia and Bipolar Disorder Working Groups ., The schizophrenia ( SCZ ) GWAS summary statistics results were obtained from the PGC Schizophrenia Work Group 13 , which consisted of 9 , 394 cases with schizophrenia or schizoaffective disorder and 12 , 462 controls ( 52% screened ) from a total of 17 samples from 11 countries ., Semi-structured interviews were used by trained interviewers to collect clinical information , and operational criteria were used to establish diagnosis ., The quality of phenotypic data was verified by a systematic review of data collection methods and procedures at each site , and only studies that fulfilled these criteria were included ., Controls were selected from the same geographical and ethnic populations as cases ., For further details on sample characteristics and quality control procedures applied , please see Ripke et al . ., The bipolar disorder ( BD ) GWAS summary statistics results were obtained from the PGC Bipolar Disorder Working Group 14 , which consisted of n\u200a=\u200a16 , 731 participants , including 7481 cases and 9250 controls , from 11 studies from 7 countries ., Standardized semi-structured interviews were used by trained interviewers to collect clinical information about lifetime history of psychiatric illness and operational criteria applied to make lifetime diagnosis according to recognized classifications ., All cases have experienced pathologically relevant episodes of elevated mood ( mania or hypomania ) and meet operational criteria for a BD diagnosis ., The sample consisted of BD I ( 84% ) , BD II ( 11% ) , schizoaffective disorder bipolar type ( 4% ) , and BD NOS ( 1% ) ., Controls were selected from the same geographical and ethnic populations as cases ., For further details on sample characteristics and quality control procedures applied , please see Sklar et al . 14 ., Due to overlapping control samples in these studies , the common controls were split randomly , and divided between the two case-control analyses ., All results presented here are based on these non-overlapping control samples , with n\u200a=\u200a9379 cases and n\u200a=\u200a7736 control samples in schizophrenia , and n\u200a=\u200a6990 cases and n\u200a=\u200a4820 controls in bipolar disorder analyses ., Analyses implemented here were motivated by previously published stratified FDR methods 5 , 37 ., However , we found that stratified empirical cdfs exhibited a high degree of variability ., Instead , we computed empirical cdfs for the first phenotype conditional on nominal p-values of the second being at or below a given threshold ., These conditional empirical cdfs vary more smoothly as a function of p-value thresholds in the second ( associated ) phenotype than do empirical cdfs employing disjoint strata ., Conditional FDR estimates derived from the conditional empirical cdfs are a simple extension of Efrons Empirical Bayes FDR methods 33 ., One advantage of the model-free empirical cdf approach is the avoidance of bias in conditional FDR estimates from model misspecification ., However , there are inherent limitations to model-free approaches , especially with respect to inferring properties of the non-null distribution and , consequently , estimating power to detect non-null effects ., We present complementary model-based analyses in the Supporting Information that estimate conditional and conjunctional local false discovery rate ( fdr ) 27 ., Results presented in the Supporting Information using this model-based fdr corroborate the results of the model-free approaches presented here ., The empirical null distribution in GWAS is affected by global variance inflation due to population stratification and cryptic relatedness 43 and deflation due to over-correction of test statistics for polygenic traits by standard genomic control methods 34 ., We applied a control method leveraging only intergenic SNPs which are likely depleted for true associations ( Schork et al . , under review ) ., First , we annotated the SNPs to genic ( 5′UTR , exon , intron , 3′UTR ) and intergenic regions using information from the 1KGP ., As illustrated in Figure S1 , there is an enrichment of functional genic regions in schizophrenia compared to the intergenic SNP category ., We used intergenic SNPs because their relative depletion of associations suggests that they provide a robust estimate of true null effects a | Introduction, Results, Discussion, Materials and Methods | Several lines of evidence suggest that genome-wide association studies ( GWAS ) have the potential to explain more of the “missing heritability” of common complex phenotypes ., However , reliable methods to identify a larger proportion of single nucleotide polymorphisms ( SNPs ) that impact disease risk are currently lacking ., Here , we use a genetic pleiotropy-informed conditional false discovery rate ( FDR ) method on GWAS summary statistics data to identify new loci associated with schizophrenia ( SCZ ) and bipolar disorders ( BD ) , two highly heritable disorders with significant missing heritability ., Epidemiological and clinical evidence suggest similar disease characteristics and overlapping genes between SCZ and BD ., Here , we computed conditional Q–Q curves of data from the Psychiatric Genome Consortium ( SCZ; n\u200a=\u200a9 , 379 cases and n\u200a=\u200a7 , 736 controls; BD: n\u200a=\u200a6 , 990 cases and n\u200a=\u200a4 , 820 controls ) to show enrichment of SNPs associated with SCZ as a function of association with BD and vice versa with a corresponding reduction in FDR ., Applying the conditional FDR method , we identified 58 loci associated with SCZ and 35 loci associated with BD below the conditional FDR level of 0 . 05 ., Of these , 14 loci were associated with both SCZ and BD ( conjunction FDR ) ., Together , these findings show the feasibility of genetic pleiotropy-informed methods to improve gene discovery in SCZ and BD and indicate overlapping genetic mechanisms between these two disorders . | Genome-wide association studies ( GWAS ) have thus far identified only a small fraction of the heritability of common complex disorders , such as severe mental disorders ., We used a conditional false discovery rate approach for analysis of GWAS data , exploiting “genetic pleiotropy” to increase discovery of common gene variants associated with schizophrenia and bipolar disorders ., Leveraging the increased power from combining GWAS of two associated phenotypes , we found a striking overlap in polygenic signals , allowing for the discovery of several new common gene variants associated with bipolar disorder and schizophrenia that were not identified in the original analysis using traditional GWAS methods ., Some of the gene variants have been identified in other studies with large targeted replication samples , validating the present findings ., Our pleiotropy-informed method may be of significant importance for detecting effects that are below the traditional genome-wide significance level in GWAS , particularly in highly polygenic , complex phenotypes , such as schizophrenia and bipolar disorder , where most of the genetic signal is missing ( i . e . , “missing heritability” ) ., The findings also offer insights into mechanistic relationships between bipolar disorder and schizophrenia pathogenesis . | medicine, psychoses, mathematics, mental health, clinical research design, heredity, statistics, genetics, schizophrenia, biology, biostatistics, statistical methods, mood disorders, complex traits, psychiatry | null |
journal.pcbi.1001096 | 2,011 | Flexible Cognitive Strategies during Motor Learning | When learning a new motor skill , verbal instruction often proves useful to hasten the learning process ., For example , a new driver is instructed on the sequence of steps required to change gears when using a standard transmission ., As the skill becomes consolidated , the driver no longer requires explicit reference to these instructions ., Operating a vehicle with a stiffer or looser clutch does not generally require further instruction , but rather entails a subtle recalibration , or adaptation of the previously learned skill ., Indeed , the use of an explicit strategy may even lead to degradation in the experts performance ., Consideration of these contradictory issues brings into question the role of instructions or explicit strategies in sensorimotor learning ., The type of motor task and nature of the instruction can have varying effects on motor execution and learning 1–3 ., In the serial reaction time task ( SRT ) , participants produce a sequence of cued button presses ., If the participant is informed of the underlying sequence , learning occurs much more rapidly compared to when sequential learning arises from repeated performance 4 ., However , learning in the SRT task entails the linkage of a series of discrete actions ., Explicit instructions of the sequence structure may be viewed as a way to create a working memory representation of the series ., Many skills lack such a clear elemental partition and , as such , participants cannot easily verbalize what a successful movement entails ., For example , the pattern of forces required to move the hand in a straight line in a novel force field 5–7 would be hard to verbalize ., Various studies have examined the role of explicit strategies in tasks involving sensorimotor adaptation 8–11 ., The benefits of an explicit strategy may be illusory with adaptive processes arising from automatic and incremental updating of a motor system that is impenetrable to conscious intervention 12–16 ., However , performance measures indicate that adaptation may differ between conditions in which participants are either aware or unaware of the changes in the environment 17 ., For example , a large visuomotor rotation can be introduced abruptly , in which case , awareness is likely , or introduced incrementally such that participants are unaware of the rotation ., The abrupt onset of large unexpected errors may promote the use of cognitive strategies 18–20 ., Participants who gain explicit knowledge of an imposed visuomotor rotation show better performance during learning than participants who report little or no awareness of the rotation 10 ., Moreover , the rate of learning , at least in the early phase of adaptation , correlates positively with spatial working memory span 21 , suggesting that strategic compensation may be dependent on working memory capacity ., Studies of sensorimotor adaptation during aging also indicate that the rate of learning is slower in older adults compared to young adults , despite similar aftereffects 22–24 ., This cost is absent in older adults who report awareness of the rotation 25 ., In many of the studies cited above , the assumption has been that the development of awareness can lead to the utilization of compensatory strategies ., However , few studies have directly sought to manipulate strategic control during sensorimotor adaptation ., One striking exception is a study by Mazzoni and Krakauer 9 ., Participants viewed a display of eight small circles , or visual landmarks , that were evenly spaced by 45° to form a large , implicit ring ., The target location was specified by presenting a bullseye within one of the eight circles ., After an initial training phase in which the visuomotor mapping was unaltered , a 45° rotation in the counterclockwise direction ( CCW ) was introduced ., In the standard condition in which no instructions were provided , participants gradually reduced endpoint error by altering their movement heading in the clockwise direction ( CW ) ., In the strategy condition , participants were given explicit instructions to move to the circle located 45° clockwise to the target ., This strategy enabled these participants to immediately eliminate all error ., However , as training continued , the participants progressively increased their movement heading in the clockwise direction ., As such , the endpoint location of the feedback cursor drifted further from the actual target location and , thus , performance showed an increase in error over training , a rather counterintuitive result 26 ., Mazzoni and Krakauer 9 proposed that this drift arises from the implicit adaptation of an internal forward model ., Importantly , the error signal for this learning process is not based on difference between the observed visual feedback and target location ., Rather , it is based on the difference between the observed visual feedback and strategic aiming location ., Even though participants aim to a clockwise location of the target ( as instructed ) , the motor system experiences a mismatch between the predicted state and the visual feedback ., This mismatch defines an error signal that is used to recalibrate the internal model ., Reducing the mismatch results in an adjustment of the internal model such that the next movement will be even further in the clockwise direction ., Thus , the operation of an implicit learning process that is impervious to the strategy produces the paradoxical deterioration in performance over time ., In the present paper , we start by asking how this hypothesis could be formalized in a computational model of motor learning ., State space modeling techniques have successfully described adaptation and generalization during motor learning 27–29 ., These models focus on how learning mechanisms minimize error from trial to trial ., Variants of these models postulate multiple learning mechanisms that operate at different time scales 28 ., Within this framework , strategic factors might be associated with fast learning processes that rapidly reduce error ., However , such models are unable to account for the drift that arises following the deployment of a strategy ., To address these issues , we developed a series of setpoint state-space models of adaptation to quantitatively explore how strategic control and implicit adaptation interact ., Assuming a fixed strategy , adaptation should continue to occur until the error signal , the difference between the feedback location and the aiming location is zero; that is , the visual feedback matches the intended aim of the reach ., As such , drift arising from implicit adaptation should continue to rise until it offsets the adopted strategy ., To test this prediction , we increased the length of the adaptation phase ., Moreover , we manipulated the salience of the visual landmarks used to support the strategy ., We hypothesized that these landmarks served as a proxy for the aiming location ., If this assumption is correct , then elimination of the visual landmarks should weaken the error signal , given uncertainty concerning the aiming location , and drift should be attenuated ., We test this prediction by comparing performance with and without visual landmarks ., When informed of an appropriate strategy that will compensate for the rotation , participants immediately counteract the rotation and show on-target accuracy ., The standard model as formulated above does not provide a mechanism to implement an explicit strategy ., To allow immediate implementation of the strategy , we postulate that there is direct feedthrough of the strategy ( s ) to the target error equation ( equation 1 ) : ( 3 ) Direct feedthrough allows the strategy to contribute to the target error equation without directly influencing the updating of the internal model ., If the strategy operated through the internal model , then the impact of the strategy would take time to evolve , assuming there is substantial memory of the internal models estimation of the rotation ( i . e . , A has a high value in Eq . 2 ) ., With direct feedthrough , the implementation of an appropriate strategy can immediately compensate for the rotation ., In the current arrangement , the appropriate strategy is fixed at 45° in the CW direction from the cued target ., Once the strategy is implemented , performance should remain stable since the error term is small ., Indeed , a model based on Eq ., 3 immediately compensates for the rotation ., The target error , the difference between the feedback location and target location , is essentially zero on the first trial with the strategy , and remains so throughout the rotation block ( Figure 1B – green line ) ., However , this model fails to match the empirical results observed by Mazzoni and Krakauer 9: performance drifts over time with an increase in errors in the direction of the strategy ., This phenomenon led the authors to suggest that the prediction error signal to the internal model is not based on target error ., Instead , the error signal should be defined by the difference between the feedback location and aiming location ( see Figure 2E ) : ( 4 ) The formulation of the prediction error term in Eq ., 4 is akin to a setpoint or reference signal from engineering control theory 30 ., In typical motor learning studies , the setpoint is to reach to the target ., When there is no strategy ( s\u200a=\u200a0 ) , the target error in Eq ., 1 is the same as the error term in Eq ., 4 ., However , when a strategy with direct feedthrough is used ( s≠0 ) , the strategy terms may cancel out if the actual implemented strategy is similar to the desired strategy ., The input error to update the internal models estimate of the rotation becomes: ( 5 ) This model shows immediate compensation for the visuomotor rotation , and more importantly , produces a gradual deterioration in performance over the course of continued training with the reaching error drifting in the direction of the strategy ( Figure 1A – red line ) , consistent with the results reported by Mazzoni and Krakauer 9 ., It is important to emphasize that the error signal for sensorimotor recalibration in Eq ., 4 is not based on the difference between the feedback location and target location ( target error ) ., Rather , the error signal is defined by the difference between the feedback location and aiming location , or what we will refer to as aiming error ., When a fixed strategy is adopted throughout training ( Figure 1B – blue line ) , the aiming error is ( initially ) quite large given that the predicted hand location is far from the location of visual feedback , even though the feedback cursor may be close to the actual target ., In its simplest form , the setpoint model predicts that , as the internal model minimizes this error ( Figure 1B – red line ) , drift will continue until the observed feedback of the hand matches the aiming location ., That is , the magnitude of the drift should equal the size of the strategic adjustment ., In the Mazzoni and Krakauer set-up 9 , the drift would eventually reach 45° in the CW direction ( Figure 1A – red line ) ., A second prediction can be derived by considering that the error signal in Eq ., 4 relies on an accurate estimate of the strategic aiming location ., We assume that a visual landmark in the display can be used as a reference point for strategy implementation ( e . g . , the blue circle adjacent to the target ) ., This landmark can serve as a proxy for the aiming location ., The salience of this landmark provides an accurate estimate of the aiming location and , from Eq ., 4 , drift should be pronounced ., However , if these landmarks are not available , then the estimate of the aiming location will be less certain ., Previous studies have shown that adaptation is attenuated when sensory feedback is noisy 31 , 32 ., One approach for modeling the effect of changing the availability or certainty of the ( strategy defined ) aiming location would be to vary the adaptation rate ( B ) ., For example , B could be smaller if there is a decrease in certainty of the aiming location , and correspondingly , a decrease in the certainty of the aiming error ., This model predicts that the rate of drift is directly related to B: if B is lower due to decreased certainty of the aiming location , then the rate of drift will be attenuated ( Figure 1A – cyan line ) ., To evaluate the predictions of this setpoint model , participants were tested in an extended visuomotor rotation task in which we varied the visual displays used to define the target and strategic landmarks ( see methods ) ., The target was defined as a green circle , appearing at one of eight possible locations , separated by 45° ( Figure 2 , only three shown here ) ., By encouraging the participants to make movements that “sliced” through the target , and only providing feedback at the point of intersection with the virtual target ring , we were able to train the participants to move quickly with relatively low trial-to-trial variability ., We assume that participants mostly relied on feedforward control given the ballistic nature of the movements and absence of continuous online feedback ., Participants were assigned to one of three experimental groups ( n\u200a=\u200a10 per group ) , with the groups defined by our manipulation of the blue landmarks in the visual displays ., For the aiming-target group ( AT ) , the blue circles were always visible , similar to the method used by Mazzoni and Krakauer ., For the disappearing aiming-target group ( AT ) , the blue circles were visible at the start of the trial and disappeared when the movement was initiated ., For the no aiming-target group ( NoAT ) , the blue landmarks were not included in the display ., The participants were initially required to reach to the green target ( Figure 2A ) ., Movement duration , measured when the hand crossed the target ring , averaged 275±50 . 8 ms with no significant difference between groups ( F2 , 27\u200a=\u200a1 . 02 , p\u200a=\u200a0 . 37 ) ., Following the initial familiarization block , participants were trained to use a strategy of moving 45° in the CW direction from the green target location , ( Figure 2B ) ., This location corresponded to the position of the neighboring blue circle ., Feedback was veridical in this phase ( e . g . , corresponded to hand position ) ., To help participants in the NoAT group learn to move at 45° , the blue circles were also presented on half of the trials for this group ( in this phase only ) ., The mean angular shifts , relative to the green target , were 43 . 4±1 . 6° and 42 . 9±1 . 2° for the AT and DAT groups , respectively ( Figure 3 - orange ) ., For the NoAT group , the mean angular shift was 43 . 5±0 . 9° when the aiming target was present and 40 . 1±7 . 1° when the aiming target was absent ., While the variance was considerably larger for trials without the aiming target , the means were not significantly different ( t18\u200a=\u200a0 . 95 , p\u200a=\u200a0 . 38 ) ., Practicing the 45° CW strategy did not produce interference on a subsequent baseline block in which participants were again instructed to reach to the cued , green target ( Figure 3 – black ) ., Over the last 10 movements of the familiarization block participants , across all groups , had an average target error of −1 . 5±0 . 7° ., Over the first 10 movements of the baseline block , this value was −0 . 5±0 . 6° , confirming that the strategy-only block did not produce a substantial bias ., Without warning , the CCW rotation was introduced ( Figure 2C ) ., As expected , the introduction of the CCW rotation induced a large target error ., Averaged over the two , rotation probe trials , the mean values were −41 . 6±3 . 3° , −43 . 8±1 . 1° , −43 . 5±3 . 2° for AT , DAT , and NoAT groups , respectively ( Figure 3 – “x” ) ., After the participants were instructed to use the clockwise strategy ( Figure 2D ) , the target error was reduced immediately to 3 . 5±4 . 4° , 1 . 0±4 . 3° , and −2 . 5±6 . 6° , values that were not significantly different from each other ( F2 , 27\u200a=\u200a1 . 96 , p\u200a=\u200a0 . 16 ) ., The participants were then instructed to use the strategy and required to produce a total of 320 reaching movements under the CCW rotation ., This extended phase allowed us to, a ) verify that error increased over time , drifting in the direction of the strategy , and, b ) determine if the magnitude of the drift would approximate the magnitude of the rotation , a prediction of the simplest form of the setpoint model ., Consistent with the results of Mazzoni and Krakauer 9 , error increased in the direction of the strategy over the initial phase of the rotation block ., However , the extent of the drift fell far short of the magnitude of the rotation ., To quantify the peak drift , each participants time series of endpoint errors was averaged over 10 movements and we identified the bin with the largest error ., Based on this estimate of peak drift , a significant difference was observed between groups ( F2 , 27\u200a=\u200a21 . 9 , p<0 . 001; Figure 4A ) ., This is consistent with the prediction of the model based that the salience of the aiming targets would influence the estimation of the aiming location ., Drift was largest when the aiming targets were always visible , and progressively less for the DAT and NoAT groups ., Drift was not isolated to particular target locations ( Figure 4B ) ., Our rotation plus strategy block lasted 320 trials , nearly four times the number of trials used by Mazzoni and Krakauer 9 ., This larger window provides an interesting probe on learning given that the participants become progressively worse in performance with respect to the target over the drift phase ., While the AT group had the largest drift , they eventually showed a change in performance such that the heading angle at the end of the rotation block was close to 45° CW from the green target ( Figure 3A ) ., By the end of training , their target error was only 0 . 3±3 . 9° , which was not significantly different from zero ( t9\u200a=\u200a0 . 17 , p\u200a=\u200a0 . 85 ) ., We did not observe a consistent pattern in how these participants counteracted the drift ( Figure 5 ) ., Two participants showed clear evidence of an abrupt change in their performance , suggesting a discrete change in their aiming strategy ., For the other eight AT participants , the changes in performance were more gradual ., The drift persisted over the 320 trials of the rotation block for participants in the DAT group ( Figure 3C ) ., The average drift was 5 . 9±4 . 8° at the end of training , a value that was significantly greater than zero ( t9\u200a=\u200a2 . 40 , p\u200a=\u200a0 . 04 ) ., Given that the NoAT group showed minimal drift , we did not observe any consistent changes in performance over the block ., At the end of training , the mean target error was only 1 . 0±1 . 9° , a value which is not significantly different from zero ( t9\u200a=\u200a1 . 01 , p\u200a=\u200a0 . 33 ) ., The availability or certainty in the estimate of the aiming location was manipulated by altering the presence of the aiming target across the groups ., As predicted by the setpoint model , the degree of drift was attenuated as the availability of the aiming targets decreased ., In the current implementation of our model , this decrease in drift rate is captured by a decrease in the adaptation rate ( B ) : with greater uncertainty , the weight given to the error term for updating the internal model is reduced ., However , one prediction of this model is at odds with the empirical results ., Variation in the adaptation rate not only predicts a change in drift rate , but also predicts a change in the washout period ., Specifically , decreasing the adaptation rate should produce a slower washout , or extended aftereffect ( Figure 1B – cyan ) ., This prediction was not supported ., The washout rates are similar across the three groups ( bootstrap , p>0 . 11 between all groups ) ., One could hypothesize different adaptation rates during the rotation and washout phases , with the effect of target certainty only relevant for the former ., However , a post hoc hypothesis along these lines is hard to justify ., Alternatively , it is possible that the adaptation rate ( B ) is similar for the three groups and that the variation in drift rate arises from another process ., One possibility is that the manipulation of the availability of the aiming targets influences the certainty of the desired strategy term in Equation 4 , and correspondingly , modifies the aiming error term: ( 6 ) A value of K that is less than 1 will attenuate drift ( Figure 1A – magenta line; simulated with K\u200a=\u200a0 . 5 ) because the strategy output ( Eq . 3 ) and the desired strategy ( Eq . 6 ) do not completely cancel out ., Consequently , the error used to adjust the internal model will be smaller and produce attenuated drift ( Figure 1B – magenta line ) ., Moreover , because the strategy is no longer used during the washout phase , the K term is no longer relevant ., Thus , the washout rates should be identical across the three groups , assuming a constant value of B . In sum , while variation in B or K can capture the group differences in drift rate , only the latter accounts for the similar rates of washout observed across groups ., When the availability of the aiming targets is reduced , either by flashing them briefly or eliminating them entirely , the participants certainty of the aiming location is attenuated ., This hypothesis is consistent with the notion that the aiming locations serve as a proxy for the predicted aiming location ., As noted above , none of the participants showed drift approaching 45° ., Even those exhibiting the largest drift eventually reversed direction such that they became more accurate over time in terms of reducing endpoint error with respect to the target location ., To capture this feature of the results , we considered how participants might vary their strategy over time as performance deteriorates ., It is reasonable to assume that the participant may recognize that the adopted strategy should be modified to offset the rising error ., One salient signal that could be used to adjust the strategy is the target error , the difference between the target location and the visual feedback ., To capture this idea , we modified the setpoint model , setting the strategy as a function of target error ( Figure 2E ) : ( 7 ) where E defines the retention of the state of the strategy and the F defines the rate of strategic adjustment ., As target error grows ( i . e . , drift ) , the strategy will be adjusted to minimize this error ( Figure 2E ) ., In our initial implementation of the setpoint model , the strategy term was fixed at 45° ., Equation 7 allows the strategy term to vary , taking on any value between 0° and 360° ., The availability of the aiming targets , captured by K in Eq ., 6 , influences the magnitude of the drift ., Greater drift occurs when the aiming error , that between the feedback location and aiming location , is salient ( Figure 1B – red line; K\u200a=\u200a1 ) ., However , when the target error grows too large , adjustments to the strategy begin to gain momentum and performance becomes more accurate with respect to the target given the change in strategy ( Figure 1C – blue line; simulated with E\u200a=\u200a1 and F\u200a=\u200a0 . 01 ) ., More emphasis on target errors rather than the aiming error results in less drift ( Figure 1C – orange line; simulated with F\u200a=\u200a0 . 05 ) ., Thus , the relative values of K and F determine the degree of performance error that is tolerated before strategic adjustments compensate to offset the drift ( Figure 1C ) ., This setpoint model ( Eqs . 8–11 ) was fit by bootstrapping ( see methods ) each groups time series of target errors: ( 8 ) ( 9 ) ( 10 ) ( 11 ) The fits ( Figure 6A–C and Table 1 ) show that K is the greatest for the AT group and progressively less for the DAT group and the NoAT group ( AT vs DAT group: p\u200a=\u200a0 . 003; AT vs NoAT groups: p<0 . 001; DAT vs . NoAT groups: p<0 . 001 ) ., When the aiming targets remain visible , the aiming error signal is readily available , and the weight given to the strategic aiming location , K , is larger ., Conversely , the weight given to the target error , F , is significantly greater for the NoAT group compared to the AT and DAT groups ( NoAT vs AT group: p\u200a=\u200a0 . 005; NoAT vs DAT group p\u200a=\u200a0 . 001 ) ., These results are consistent with the hypothesis that participants in the NoAT group rely more on target errors because the absence of the aiming targets removes a reference point for generating a reliable aiming error ( Eq . 9 ) ., The dynamics of the recalibration process and strategy state ( Eqs . 10 and 11 ) are plotted in Figure 6D ., These parameters , along with the other parameters that represent the memory of the internal model ( A ) , the adaptation rate ( B ) , and the memory of the strategy ( E ) are listed in Table 1 ., Following the rotation block , we instructed the participants that the rotation would be turned off and they should reach to the cued green target ., For the first eight trials , no endpoint feedback was presented ., This provided a measure of the degree of sensorimotor recalibration in the absence of learning ( Figure 4C – triangles ) ., Aftereffects were observed in all three groups ., The average error was significantly different from zero in the CW direction from the green target for all three groups ( one sample t-test for each group , p<0 . 001 ) ., In comparisons between the groups , the AT group showed the largest aftereffect of 19 . 2±3 . 7° ( t18\u200a=\u200a3 . 5 , p\u200a=\u200a0 . 003 and t18\u200a=\u200a5 . 61 , p<0 . 001 compared to the DAT and NoAT groups , respectively ) ., The mean aftereffects for the DAT and NoAT groups were 10 . 4±3 . 2° and 6 . 8±2 . 2° , values that were not significantly different ., When endpoint feedback was again provided , the size of the aftereffect diminished over the course of the washout block ( Figure 4C - squares ) ., In the setpoint model , the internal model will continue to adapt even in the face of strategic adjustments adopted to improve endpoint accuracy ., As such , the model predicts that the size of the aftereffect should be larger than the degree of drift ., To test this prediction , we compared the peak drift during the rotation block to the aftereffect ., In the preceding analysis , we had estimated peak drift for each participant by averaging over 10 movements and identifying the bin with the largest error ., However , a few errant movements could easily bias the estimate of drift within a 10-movement bin ., As an alternative procedure , we used a bootstrapping procedure to identify the bin with the largest angular error for each group ., This method should decrease the effect of noise because the estimate of peak drift is selected from an averaged sample of the participants data ., Moreover , any bias in the estimate of the magnitude of the peak should be uniform across the three groups of participants ., For consistency , we estimated the aftereffect ( the first 8 trials without feedback ) using the same bootstrap procedure ., For the AT group , the peak drift was 14 . 8±2 . 5° in the CW direction , occurring 64±30 movements into the rotation block ., For the DAT group , the peak drift was 10 . 0±1 . 8° , occurring at a later point in the rotation block ( 130±106 ) ., For the NoAT group , peak drift was only 3 . 2±2 . 7° and occurred after 145±131 movements ., As predicted by the model , the aftereffect was significantly larger than peak drift for the AT and NoAT groups ( Figure 4D; bootstrap: p\u200a=\u200a0 . 002 and p<0 . 001 , respectively ) ., The difference between the degree of peak drift and aftereffect in the DAT group was not reliable ., It is important to emphasize that estimates of the time of peak drift should be viewed cautiously , especially in terms of comparisons between the three groups ., These estimates have lower variance for the AT group because it was easier to detect the point of peak drift in this group compared to the DAT and NoAT groups ., Visuomotor rotation tasks are well-suited to explore how explicit cognitive strategies influence sensorimotor adaptation ., Following the approach introduced by Mazzoni and Krakauer 9 , we instructed participants to aim 45° CW in order to offset a −45° rotation ., Between groups , we manipulated the information available to support the strategy by either constantly providing an aiming target , blanking the aiming target at movement initiation , or never providing an aiming target ., In all groups , the strategy was initially effective , resulting in the rapid elimination of the rotation-induced endpoint error ., However , when the aiming target was present , participants showed a drift in the direction of the strategy , replicating the behavior observed in Mazzoni and Krakauer 9 ., This effect was markedly attenuated when the aiming target was not present suggesting that an accurate estimate of the strategic aiming location is responsible for causing the drift ., In addition , when the drift became quite large ( as in the AT group ) , participants begin to adjust their strategy to offset the implicit drift ., Mathematical models of sensorimotor adaptation have not explicitly addressed how a strategy influences learning and performance ., By formalizing the effect of strategy usage into the standard state-space model of motor learning , we can begin to evaluate qualitative hypotheses that have been offered to account for the influence of strategies on motor learning ., Mazzoni and Krakauer 9 suggested that drift reflects the interaction of the independent contribution of strategic and implicit learning processes in movement execution ., Current models of adaptation cannot be readily modified to account for this interaction ., Rather , we had to consider more substantive architectural changes ., Borrowing from engineering control theory , we used a setpoint model in which the internal model can be recalibrated around any given reach location ., The idea of a setpoint is generally implicit in most models of learning , but this component does not come into play since the regression is around zero ., However , simply making the setpoint explicit is not sufficient to capture the drift phenomenon ., The strategy must have direct feedthrough to the output equation in order to implement the explicit strategy while allowing for an internal model to implicitly learn the visuomotor rotation ., This simple setpoint model was capable of completely eliminating error on the first trial and capture the deterioration of performance with increased training ., Drift arises because the error signal is driven by the difference between the internal models prediction of the aiming location and the actual , endpoint feedback ., The idea that an aiming error signal is the source of drift is consistent with the conjecture of Mazzoni and Krakauer 9 ., An important observation in the current study is that , given uncertainty in the prediction of the aiming location , participants use external cues as a proxy in generating this prediction ., This hypothesis accounts for the observation that drift was largest when the aiming target was always visible , intermediate when the aiming target was only visible at the start of the trial , and negligible when the aiming target was never visible ., The aiming target , when present , served as a proxy for predicted hand position , and helped define the error between the feedback cursor and aiming location in visual coordinates ., When the aiming target was not present , the aiming location was less well-defined in visual coordinates , and thus , the relationship between the aiming location and feedback cursor was less certain ., Under this condition , the participants certainty of the error was reduced and adaptation based of this signal was attenuated ., Quantitatively , progress | Introduction, Results, Discussion, Methods | Visuomotor rotation tasks have proven to be a powerful tool to study adaptation of the motor system ., While adaptation in such tasks is seemingly automatic and incremental , participants may gain knowledge of the perturbation and invoke a compensatory strategy ., When provided with an explicit strategy to counteract a rotation , participants are initially very accurate , even without on-line feedback ., Surprisingly , with further testing , the angle of their reaching movements drifts in the direction of the strategy , producing an increase in endpoint errors ., This drift is attributed to the gradual adaptation of an internal model that operates independently from the strategy , even at the cost of task accuracy ., Here we identify constraints that influence this process , allowing us to explore models of the interaction between strategic and implicit changes during visuomotor adaptation ., When the adaptation phase was extended , participants eventually modified their strategy to offset the rise in endpoint errors ., Moreover , when we removed visual markers that provided external landmarks to support a strategy , the degree of drift was sharply attenuated ., These effects are accounted for by a setpoint state-space model in which a strategy is flexibly adjusted to offset performance errors arising from the implicit adaptation of an internal model ., More generally , these results suggest that strategic processes may operate in many studies of visuomotor adaptation , with participants arriving at a synergy between a strategic plan and the effects of sensorimotor adaptation . | Motor learning has been modeled as an implicit process in which an error , signaling the difference between the predicted and actual outcome is used to modify a model of the actor-environment interaction ., This process is assumed to operate automatically and implicitly ., However , people can employ cognitive strategies to improve performance ., It has recently been shown that when implicit and explicit processes are put in opposition , the operation of motor learning mechanisms will offset the advantages conferred by a strategy and eventually , performance deteriorates ., We present a computational model of the interplay of these processes ., A key insight of the model is that implicit and explicit learning mechanisms operate on different error signals ., Consistent with previous models of sensorimotor adaptation , implicit learning is driven by an error reflecting the difference between the predicted and actual feedback for that movement ., In contrast , explicit learning is driven by an error based on the difference between the feedback and target location of the movement , a signal that directly reflects task performance ., Empirically , we demonstrate constraints on these two error signals ., Taken together , the modeling and empirical results suggest that the benefits of a cognitive strategy may lie hidden in many motor learning tasks . | neuroscience/behavioral neuroscience, neuroscience/cognitive neuroscience, neuroscience/motor systems, computational biology/computational neuroscience, neuroscience/experimental psychology | null |
journal.ppat.1004737 | 2,015 | A Temporal Gate for Viral Enhancers to Co-opt Toll-Like-Receptor Transcriptional Activation Pathways upon Acute Infection | Infection by pathogens is detected by the host innate immune system through interaction of Pathogen-Associated Molecular Patterns ( PAMPs ) using a range of extra and intra-cellular host Pathogen-Recognitions-Receptors ( PRRs ) 1–3 ., The major group of PRRs is represented by the family of Toll-Like-Receptors ( TLRs ) that detect a range of PAMPs and are located either at the cell surface , e . g . TLR2 and TLR4 , or in endosomes , e . g . TLR3 , 7 and 9 3–5 ., Binding of the corresponding ligands to these receptors leads to the activation of downstream signalling factors and TLR-receptors are dependent on the adaptor molecule MyD88 , with exception of TLR3 and 4 ., TLR3 signals exclusively through the adaptor TRIF and TLR4 is the only TLR that can utilise both signalling pathways 3 , 6 ., The activity of the TLR-signalling pathway triggers the expression of type I interferons and other antiviral factors that aid to control infections 7–9 ., Cytomegalovirus ( CMV ) is recognised by the innate immune system using a diverse set of PRRs 1 , 10 , 11 ., At the cell surface a direct interaction between the viral glycoproteins and TLR2 has been reported for human CMV ( HCMV ) 12 , 13 and also for the related human Herpesvirus 1 ( HSV1 ) 14 ., Other TLRs that play a role in resistance to CMV infection are TLR3 and TLR9 ., Homozygotic knockout animals for Tlr2 , Tlr3 or Tlr9 are highly susceptible for CMV infection and show increased mortality rates 15 , 16 ., Other types of PRRs have also been implicated to contribute to the detection of CMV infection ., The cytoplasmic DNA sensors DAI ( ZBP-1 ) 17 and AIM2 18 have been shown to detect CMV ., The interaction of viruses at the cell surface ( TLR2 ) , or intracellular recognition of viral genomes ( by DAI , AIM2 , TLR9 ) and virion packaged RNA ( through RIG-I , TLR3 , TLR7 ) 19 results in triggering anti-viral responses through the signal activation of downstream inflammatory transcription factors ( TFs ) ., Depending on the infected cell type , these signal regulated TFs include NFκB , AP1 , CREB/ATF , IRF3 or IRF7 which govern the expression of pro-inflammatory and anti-viral host factors and effector molecules ., Virus genomes encode a number of proteins , termed evasins that help to evade and subvert the host immune response to the infection 20–23 , many of which target molecules of the adaptive immune response ., Some evasins , however , inhibit the innate immune response to infection , in particular the production of IFN ., For example the UL83 gene product pp65 24 , 25 and the IE86 protein ( IE2 ) of HCMV and the Ie1 protein of MCMV have been reported to moderate production of pro-inflammatory cytokines 26 , 27 ., Human CMV has also been shown to disrupt functionality of the interferon stimulated gene factor 3 ( ISGF3 ) , reducing IFNα production 28 and very recently the early gene UL26 has also been described to antagonise NFκB activation 29 ., Two well-characterised inhibitors of innate immune signalling in murine CMV ( MCMV ) infection are the proteins M27 and M45 ., M27 and its HCMV homologue UL27 are efficient inhibitors of Type I and Type II IFN signalling through interaction and degradation of STAT2 and interference with tyrosine phosphorylation 30–32 , therefore interfering with downstream autocrine and paracrine effects of TLR activation with the exception of plasmocytoid dendritic cells 33 ., On the other hand , expression of M45 during the early phase of the infection cycle has been demonstrated to block NFκB activation , therefore interfering directly with PRR signalling pathways 34 ., The mechanism of action for M45 is based on interaction with RIP1/3 and NEMO , proteins involved in the signalling cascade controlling the degradation of the inhibitor of NFκB , Ikbα 34 , 35 ., However , de novo expression of both inhibitors M27 30–32 and M45 36 is necessary for their inhibitory activity and takes place during the early phase of the infection cycle ., Of these proteins only M45 has been detected in virions 37 ., Recently and unexpectedly the viral particle associated M45 protein has been shown to promote the activation of NFκB in fibroblasts during the immediate early ( IE ) phase of infection 36 ., However , the functional relevance of this to infection is not clear at present ., This poses the question if other mechanisms are in place to ensure sufficient viral gene expression despite the activation of anti-viral signalling events during the IE-phase of the CMV transcription-replication cycle ., IE-gene expression is under control of a potent enhancer that plays a critical role in determining success of a productive CMV infection 38–42 ., In vivo the loss of the complete enhancer results in greater than a 3-log reduction in viral load and fails in exponential growth 31 , 34 ., Indeed , the human CMV genome has been long established to contain one of the strongest known enhancers as part of its major immediate early promoter ( MIEP ) with a 650 bp core that binds multiple transcription factors and which governs expression of the viral IE-genes 43–45 ., While this region is functionally present in all CMV genomes the enhancer sequences are not conserved but instead share many of the same regulatory TF binding elements 41 ., In particular all CMV enhancer regions contain a large number of highly redundant signal-regulated transcription factor binding sites , such as those interacting with NFκB , AP1 and CREB/ATF , factors that can be also activated by the TLR signalling pathways , 5 , 41 , 46 , 47 ., This overlap combined with the combinatorial flexibility of the enhancer TF interactions indicates a potential for CMV to utilise the activation of anti-viral signalling pathways in the host cell ., It has been reported that TLR9 stimulation plays both positive and negative roles in HCMV infection 48 and that TLR4 and TLR9 activation can increase gene expression from a human CMV enhancer expression plasmid 46 ., It is thus conceivable that the CMV enhancer might advantageously co-opt the triggered TLR-signalling pathway and therefore efficiently initiate its transcription-replication-cycle before the host-cell could produce any anti-viral effector molecules 47 ., The basis of this concept has been discussed before 49 however , previous models have focused mainly on the role of NFκB , placing hijacking of NFκB signalling at the centre of the co-opting strategy ., Notably , for both HCMV and MCMV it has been shown that NFκB is not essential for wild-type virus to drive its gene expression and only becomes crucial when other TF binding sites are impaired 50 , 51 ., This poses the question whether the CMV viral enhancer has evolved a functional role in effectively co-opting multiple redundant immune signal-regulated TFs for initiating a productive transcription-replication cycle ., Hence , the underlying hypothesis for the present study is that inflammatory signalling at immediate-early times may promote viral infection through viral enhancer sequences ., We report our first experimental tests to refute this hypothesis by systematically investigating the requirements and mechanisms for innate immune regulation of the CMV enhancer , in particular upon infection of macrophages and upon in vivo infection ., We use a combination of RNAi library screens with host and viral genetics to delineate the TF network controlling the enhancer ., Our findings reveal an integrated inflammatory TF-network consisting of IRF5 , SP1 , RXR and NFκB pathways with signal activation strongly dependent on MyD88 that is delimited by a specific temporal window for activation ., These results support the hypothesis and further advance the concept of viral enhancer mimicry of innate immune promoters as an immune evasion strategy ., We first sought to examine whether viral IE-genes show similar expression kinetics to host innate immune genes upon infection of macrophages and test if they react to common stimuli ., For these experiments we compared the expression kinetics of the host innate immune genes Ifnb1 , Il6 and Tnf with the viral major IE-gene M123 ( Ie1 ) by relative qPCR ., As expected infection of bone marrow derived macrophage ( BMDM ) cells from C57/BL6 mice with MCMV triggered expression of the host pro-inflammatory cytokines IFNβ , IL6 and TNF ( Fig . 1A ) ., Notably , the overall kinetics of the viral Ie1 gene and the host factors are similar , with a rapid induction of gene expression within the first 2 h post infection ( hpi ) ., To more extensively examine the induction of host innate immune genes we performed a microarray clustering analysis of expression levels in BMDMs after MCMV infection ( Fig . 1B ) ., For this study we used a set of well-known innate immune genes and TLR signalling components ., This analysis revealed genes with expression profiles similar to the Ie1 expression profile shown in Fig . 1A , with a rapid induction within 2–4 hpi ( Fig . 1B , left panel ) ., Exceptions of this pattern were Il10 , Tlr4 , Tlr7 and Tlr9 ., Il10 , Tlr7 and 9 were induced with delayed kinetics and reaching peak expression levels by 6 h or later ., In contrast , Tlr4 seemed to have a high level of steady state expression in the mock sample and was down regulated after infection ., To determine whether these changes in gene expression can be , at least partially , triggered by TLR activation alone , we further analysed the same set of genes in BMDMs stimulated with Poly IC , which activates TLR3 signalling ( Fig . 1B , right panel ) ., The stimulation with Poly IC recapitulates the observed pattern triggered by CMV infection with the exception of TLR4 expression ., The down regulation of TLR4 observed in the infected sample was strongly delayed after Poly IC challenge ., This might reflect the synergistic effect of parallel activation of several PAMPs by the infection process , which could explain the more pronounced down regulation in the infected versus the single stimulus by Poly IC ., Since HCMV-enhancer-driven reporter plasmids have been shown to positively respond to LPS and CpG 46 , we tested if the viral IE-gene expression responds to stimulation of TLR signalling in the context of the viral infection ., For these experiments RAW264 . 7 and primary BMDMs were pre-stimulated for 15 min with ligands for TLR4 , TLR3 , TLR2 , TLR7 and TLR9 to stimulate the TFs activated by TLR signalling ., These cells were subsequently infected with a gaussia luciferase 52 , 53 reporter virus ( MCMV-gLuc ) to quantitatively measure in the context of infection the activity of enhancer ., This recombinant virus ( S1 Fig for structure and mutagenesis strategy ) has had the dispensable m128 ( Ie2 ) gene 54–56 replaced by a reporter cassette expressing a gaussia luciferase ( gluc ) reporter under direct control of the murine CMV enhancer ., Levels and kinetics of gluc and Ie1 expression in this reporter mutant are comparable as demonstrated by qPCR measurement ( S2 Fig ) ., Subsequent to TLR stimulation the cells were infected and 2 hpi the activity of the secreted gLuc reporter was measured in the cell culture supernatant ( uninfected background for BMDMs was determined as 370 . 28 RLU with SEM = 4 . 81 , n = 35 independent biological experiments; uninfected background for RAW264 . 7 was 82 RLU with SEM = 1 . 44 , n = 32 ) ., As shown in Fig . 1C ( left panel ) and Fig . 1D ( left panel ) , both cell systems showed a significant increase in reporter gene expression for specific TLR ligands ( pre-normalised average of RLU for mock in BMDMs = 3 . 23x104 RLU with min = 1 . 26x103 and max = 1 . 69x105 , n = 50; not normalised average mock for RAW264 . 7 = 4 . 04x102 RLU with min = 9 . 6x101 and max = 9 . 39x102 RLU , n = 18 ) ., Notably , there are differences between the monocytic cell line RAW264 . 7 and the primary BMDMs in the measured levels of gene expression , for TLR4 , TLR2 and TLR7 ligands LPS , Pam3CSK4 and R848 , respectively ., In these cases the RAW264 . 7 cells show a stronger response to the respective ligands , which is most likely related to differences in expression levels of their respective TLR repertoire ., We therefore compared the expression levels of selected TLRs between BMDMs and RAW264 . 7 cells and found all of the assessed TLRs had increased expression levels in the RAW264 . 7 cells ( S3 Fig ) ., We furthermore could find no significant changes in uptake of viral genomes after TLR ligand treatment in BMDMs ( S4 A and B Fig ) , indicating that the observed increase in gLuc activity is due to increased gene expression ., To check if the used TLR ligands are biologically active and can induce host innate immune gene expression , we measured induction of Ifnb1 , Il6 and Tnf in primary BMDMs and RAW264 . 7 by the same TLR ligands ( Fig . 1C and Fig . 1D , right panels ) ., Notably , Poly IC was ineffective in activating gene expression in RAW264 . 7 cells in contrast to BMDMs ., This might be due to differences in uptake of Poly IC since TLR3 is mainly localised in endosomes and its subcellular location is cell type dependent 57–60 ., Taken together these data clearly show that inflammatory stimuli that induce expression of innate immune genes also enhance viral IE-gene expression in the analysed cell systems ., Altogether we conclude that the CMV enhancer is activated with similar kinetics to innate immune genes in the context of infection and is positively responsive to inflammatory TLR-signalling ., Our results so far indicate that stimulation of TLR signalling close to the time point of infection is sufficient to increase viral enhancer activity ., While this concurs with work that demonstrated a stimulation of human CMV reporter-plasmids after LPS/CpG treatment , it is in contrast to the well-established observation that activation of TLR signalling is necessary for resistance to pathogens such as CMV 15 ., As shown in Fig . 2A , long term pre-treatment ( 24 h ) of BMDMs with all TLR ligands used in this study significantly inhibits MCMV as expected ., The inhibitory effects of the TLR-ligands are most likely due to the induced expression of anti-viral effectors such as IFNβ ( compare with Fig . 1C and 1D ) that subsequently establish an anti-viral state through autocrine and paracrine effects 61–63 ., For the anti-viral effector molecule IFNγ we have shown previously 64 , that it has a half maximum pre-treatment time ( ET50 ) of 1 . 5 h to impart 50% inhibition of CMV enhancer activity in BMDMs ., Since production of IFNs in naive cells needs to be induced first , this indicated that there should be a lag in the system between first contact with the virus and production of IFNs ., In agreement , we have previously found that upon infection of BMDMs , IFNβ secreted protein levels peak by 6 hpi 64 ., This lag predicts a possible temporal gate that would be open for a period no more than 7 hours for MCMV to establish infection before anti-viral effectors fully inhibit the virus ., Therefore we sought to determine if the anti- and pro-viral effects we observed are dependent on a time window and if so to define the boundaries of this potential temporal gate by comparing different pre-treatment times ranging from 0–24 h ., When we tested pre-treatment times with ligands for TLR2 , TLR4 and TLR9 ranging from parallel treatment ( 0 min ) to 60 min pre-treatment , we found that all conditions were pro-viral , showing that there seems to be no measurable lower limit for the temporal gate ( Fig . 2B ) ., To establish the upper limit of the temporal gate we tested pre-treatment times ranging from 1–24 h ., Fig . 2C shows that the effects of TLR ligands for TLR2 , TLR4 and TLR9 changed over time from anti-viral to pro-viral with decreasing pre-treatment times ., Matching the observations shown in Fig . 2A , the TLR ligands showed different levels of anti-viral activity with 24 h pre-treatment , with TLR2 showing only weak anti-viral activity and becoming pro-viral from 6 h or less ., The more highly potent anti-viral states induced by TLR4 and TLR9 agonists , required shorter pre-treatment times to establish resistance to infection ., To more precisely quantify the temporal window we determined the half maximal time for pro-viral enhancer stimulation by computing the ET50 for the TLR agonists ., We fitted a regression function to the linear phase of the response ( Fig . 2D ) and estimated the ET50 values as -5 h 15 min for TLR4 and -6 h 9 min for TLR9 , respectively ., Therefore , these experiments revealed the existence of a temporal gate in which CMV IE-gene expression can co-opt TLR signalling to its advantage , within the first 6 h of infection ., We next sought to investigate if the observed boost in enhancer activity conveyed a benefit for viral production ., We assessed effects of TLR signalling on viral replication by quantifying replication of the viral genome and the production of infectious particles ., To measure viral genome replication we used absolute quantification of viral genome copies by qPCR in infected BMDMs at 24 hpi , a time point at which the first round of replication is completed 65 ., Detection of the host gene Gapdh was used to correct for potential variation in the amount of input material and values were furthermore normalised to the copy numbers of the mock sample ( average genome copy number without normalisation was 1 . 73x105 with min = 3 . 28x103 copies and max = 2 . 5x105 copies ) ., Fig . 2E shows that 15 min pre-treatment with all tested TLR ligands significantly increased viral genome replication compared to the mock control ., We then tested the impact of TLR activation on the production of infectious viral particles by standard plaque assay ., To reduce interference of IFNs produced by the infected experimental BMDM cultures and to increase sensitivity we used Stat1-/- MEFs for the plaque assay ., As shown in Fig . 2F all treatments , except for Poly IC , significantly increased the production of viral particles in the culture at 3 days post infection ., While quantitative variation in IE gene expression is observed for the different ligands , the levels of boosted IE-gene expression and viral genome replication do not completely reflect the detected increase in viral particle production in the plaque assay ., Although we do not fully understand this variation it is most likely due to the downstream anti-viral factors and their effects elicited by the respective TLR signalling pathways ( compare to Fig . 1 and Fig . 2A ) ., Nevertheless these data demonstrate that TLR activation can boost at IE-times of infection both viral gene expression and subsequent replication and supports the notion that CMV enhancer might co-opt activity of innate immune signalling to its own advantage ., Comparison of TFs known to bind the HCMV enhancer with TFs activated by TLR signalling show that several factors are shared , including NFκB , AP1 and ATF 41 , 66–68 ., To analyse the importance of these TFs in our system we used a set of chimaeric viral recombinants in which the human CMV wild-type and mutant enhancers replace the native murine CMV enhancer 38 , 69 ., We analysed the effects of TLR activation on Ie1 gene expression in the chimaeric virus carrying the wild-type human CMV enhancer ( hMCMV ) and compared this with a triple knockout mutant ( hMCMV-Δ3 ) in which all enhancer binding motifs for NFκB , AP1 and ATF have been rendered non-functional by point mutations ., As can be seen in Fig . 3 the disruption of the binding motifs significantly reduced the expression of Ie1 after TLR stimulation , demonstrating that the observed effects are in part due to direct activation by NFκB , AP1 and/or ATF ., However , we note that the mutation of all NFκB , AP1 and ATF binding sites could not completely abolish the boost of viral Ie1 gene expression by TLR activation ., Stimulation of TLR4 ( p-value≤0 . 05 ) , TLR3 ( p-value≤0 . 01 ) and , with a trend , TLR2 ( p-value = 0 . 099 ) was still able to boost viral gene expression ., This result indicates that additional host factors must also be involved in the stimulation of CMV enhancer activity by TLR signalling ., It is possible that a viral tegument protein rather than indirect TLR signalling could lead to activation of TFs ., A very recent publication showed that the viral protein M45 , which is delivered into cells initially by viral particles , is a potent activator of NFκB signalling at IE times despite its role as an inhibitor of NFκB at early and late times of infection 36 ., However , these studies were limited to fibroblasts and found no necessity for M45 to activate viral IE-gene expression ., To assess whether M45 or other tegument proteins are involved in the observed TLR-mediated effects on IE-gene expression , we first examined the effect of TLR stimulation on genomic MCMV-gLuc DNA transfected into primary MEFs ., After the transfection cells were visually checked for successful transfection by assessing if individual fluorescent cells were present in the cultures ., Subsequently , half of the transfected cultures were treated with LPS to trigger TLR signalling ., Over the first 8h we checked repeatedly for changes in expression of the reporter gene gLuc and found that from 4h post treatment a shift in in a number of independent LPS treated cultures to higher reporter expression was detectable ., To determine if this initial boost of gene expression translated into increased viral replication , we assessed the number of plaques and the size of plaques at 6 days post treatment ( Fig . 3D ) ., This analysis showed that the group of LPS treated cultures produced more viral plaques in total and had statistically significant higher numbers of plaques in the small and large categories ., It is noteworthy that large plaques were mainly present in the LPS treated group ., This experiment demonstrates that an initial boost through TLR stimulation can translate into an enhancement of the transcription-replication cycle from viral genomic DNA ., This observation further supports the notion that the observed effects are not absolutely dependent on viral tegument proteins ., Since it has been recently demonstrated that in NIH3T3 fibroblasts the activation of NFκB immediately after infection is strictly dependent on M45 36 , we next assessed the role of this tegument protein in the macrophage system using a ΔM45 recombinant virus ., A comparison of protein levels in MEFs infected with wild type and mutant virus confirmed the phenotype published for NIH3T3 cells ., In MCMV infected MEFs a rapid degradation of the inhibitor of NFκB ( IκBa ) could be observed that was blocked at later times ( ≥5 hpi ) when de novo synthesised M45 became visible ( Fig . 3E , top left panel ) ., In MEFs infected with the MCMV-ΔM45 virus the degradation of IκBa only became detectable at 5 hpi ., In contrast to this , rapid degradation of IκBa was observed in the SVEC4-10 endothelial cell line and in the macrophage cell lines IC-21 and RAW264 . 7 after infection with both , the MCMV and MCMV-ΔM45 viruses ( Fig . 3E , top right and lower panels ) ., This demonstrates that , in contrast to fibroblast cells , NFκB activation at IE-times in infected endothelial and macrophages is not dependent on M45 ., Overall , the loss of M45 did not overtly impact on IE1 protein levels in any of the tested systems ., In separate experiments we also find the M45 MCMV mutant was still responsive to TLR activation in macrophages by stimulation with TLR2 agonist ( S5 B Fig ) ., We conclude from these experiments a direct involvement of the enhancer elements in mediating responsiveness to TLR-signalling and while not mutually exclusive from the possible contribution of viral particle associated proteins , is shown to be independent of the virion-delivered M45 protein ., To explore more fully the TF network required for enhancer activation , we used a library of small interfering RNAs ( siRNAs ) for targeted knock-down of immune signalling components and TFs that are either known to bind the CMV enhancer or to be activated by TLR signalling and a range of positive and negative control genes ( targeting in total 149 host factors , complete list of targets are shown in S1 Table ) ., We transiently transfected primary MEFs with targeted SMARTpools ( Dharmacon , Lifetechnologies ) and subsequently infected them with the gLuc-reporter virus ( MCMV-gLuc ) to monitor exclusively enhancer activity at 6 hpi ( 25nM siRNA screens ) ., Fig . 4A shows the summarised results of 4 independent , normalised gLuc screens that passed the quality control filter ( detectable expression of the reporter in the mock transfected samples and knock-down by several positive control siRNAs , e . g . targeting either the reporter gLuc or TFs known to be important for CMV enhancer activation , e . g . Sp1 68 , 70 ) ., Due to a high level of redundancy in the CMV enhancer overall inhibitory effects of the knockdowns rarely reach more than two-fold , even with the positive controls ., We therefore ranked the siRNA knockdowns relative to the maximal achieved knockdown by the gLuc targeting positive control siRNA ( Ranked list of all genes see S1 Table , for corresponding knock-down data see S2 Table ) ., For analysis we set cut-offs representing different ranks of our 149 targeted host factors ., We used three levels of stringency to sort our target list; the high stringency group ( >75% of maximal knock-down effect ) consisted of the top 25 candidate genes , the medium stringency group ( >50% ) included the top 63 genes and the low stringency group ( >25% ) included the top 101 gene candidates ., To investigate whether the hits from the screen were dispersed over a range of different interaction networks or were limited to a discrete network of biochemical/molecular interactions we undertook a network analysis of the high stringency group of 25 candidates with the STRING web tool ., This approach determines edge connectivity of the hits based on known and predicted molecular interactions 71 ., The results of this analysis shown in Fig . 4B , reveal that most of the target genes of the >75%-group could be mapped to a principal network with the TLR-adaptor protein MyD88 at its centre ( confidence of interaction is indicated by thickness of connecting edges , asterisks next to network node indicates statistical significant knock-down ) and a link to an RXR network ., We found that a substantial part of our statistically significant hits could be mapped to the TLR immune response pathway ( top hit in GO term enrichment test was “activation of innate immune response” with p-value≤7 . 609x10-9 ) with associated innate immune factors , such as TLR7 , IRAK1/4 , MyD88 , IRF5 and also RIG-I ( Ddx58 ) , and AIM2 ., This functional network was connected to the NFκB subunit RelA as was to be expected but also included the TFs , SP1 , ETS1 , Nfyb , Nfyc and RXRA ., The presence of RXRA , ETS1 and SP1 in the list of significant hits affecting viral enhancer activity was not surprising , since interactions with the CMV enhancer for SP1 70 , the ETS-family ( ELK1 72 , ETS-2 73 ) and RXRA 41 , 74–76 have been described in the literature ., In support , Fig . 4C shows the normalised average gLuc activity of all members of the RXR network , the IRF family and the TLR signalling components for comparison ., Notably , the factors IRF5 , AIM2 , RIG-I , Nfyb and Nfyc have not been previously implicated in mediating activation of CMV gene expression ., An analysis of the medium and low stringency factors with the topology inferred by the STRING software tool showed a highly integrated network , comprising TLR signalling components linked to the RXR network ( see S6 and S7 Figs ) ., We next sought to assess whether transcription factor requirements are convergent or divergent for viral growth and enhancer activity ., Due to the inherent redundancy of TF requirement for enhancer activity we , therefore , conducted a large number ( up to n = 24 ) of systematic independent siRNA screens for the 149 TFs ., For these screens we used a GFP-expressing reporter virus ( MCMV-GFP ) to monitor viral replication at 72 hpi and MCMV-gLuc to monitor enhancer activity at 6 hpi ., By applying a robust statistical meta-analysis we aimed to increase the statistical power and identify the most consistent siRNA effects on the results of all screens over all experimental conditions ., Fig . 5 shows the medians of each siRNA over all screens from all experimental conditions with those highlighted in bold being significantly different from the infected controls ( number of screens per siRNA up to n = 24 , see figure legend ) ., The upper panel of Fig . 5 shows replication efficiency at 72 hpi , while the lower panel shows enhancer activity at 6 hpi ( see also S3 Table ) ., This analysis allowed for statistically stringent assessment of viral replication and enhancer activation screens and revealed , as expected from the literature , components of the TLR signalling pathway , such as TAK1 , TBK1 and IRF2 that have a statistically significant anti-viral effect when measuring replication ., In contrast and in concordance with the temporal gate model for immune activation we also find many of the TLR signalling components , such as TLR9 , IRAK2 , IRAK4 and IRF3 , IRF4 , IRF5 and IRF6 to have a pro-viral effect shown in the lower panel of Fig . 5 ., For significant hits and overlap between the two screens see Table 1 ., Assessing the overlap between the two approaches , we find several genes that significantly reduced enhancer activation as well as viral replication , namely ELK1 , FOSB , IRF4 , IRF5 , IRF6 , RELA , RXRA , Sp1 , SP6 , SP7 , SRF and YY1 ., Notably , IRF5 , RELA , RXRA , SP1 and YY1 were also part of the network identified in the initial enhancer activation screen ( Fig . 4 ) ., Taken together these targets indicate that discrete innate immune signalling plays an important role in activating the viral enhancer ., Overall this data implies that the TLR-activated host factors NFκB ( RelA ) , SP1 , RXR and members of the IRF family play a central role in activating the enhancer in infection and that these factors may form a functional network ., Thus , the TLR signalling pathway might be necessary for normal IE-gene expression levels in infections with the potential for cooperation with the retinoic-acid signalling pathway ., The above-described screens identified RXRA as part of the integrated TLR-network affecting viral IE-gene expression ., While retinoic acid receptors have been shown to bind to and regulate human and murine CMV enhancer activity 74 , 77 , it is known that retinoic acid receptors can also positively influence TLR expression 78 , 79 ., To functionally test the effects of retinoids in our system , we pre-treated BMDMs with the RAR/RXR ligand 80 9-cis-retinoic acid ( 9-cis-RA ) for 24 h prior to TLR-pre-treatment ., As shown in Fig . 6A , we find that the 9-cis-RA pre-treatment triggers an increased sensitivity of our cells to the effects of TLR pre-treatment ., The ratios of gLuc activity between 9-cis-RA-treated and vehicle-treated samples show that RA has a broadly positive effect on the system , since all TLR treatments , independent of the ligand , showed ratios significantly larger than 1 ( Fig . 6B ) ., This suggests that the viral enhancer may also directly benefit from RA stimulation ., As can be seen in Fig . 6C , we monitored gLuc expression in vehicle and 9-cis-RA treated BMDMs over a time course and observed that the 9-cis-RA treatment increases IE-gene expression for up to 72 hpi in absence of any prior TLR ligand treatment ., The direct effects of RA on viral gene expression are mediated through multiple high affinity retinoic acid receptor binding sites ( RA Response Elements , RAREs ) that have been previously characterised for both the HCMV and MCMV enhancers and shown to influence IE-gene expression 74–76 ., To analyse if the 9-cis-RA treatment has a direct effect we used a chimaeric murine CMV mutant similar to those described above , in which all RAREs in the human CMV enhancer have been disrupted by point mutations ( hMCMV-ΔRARE ) and measured Ie1 expression by qPCR . | Introduction, Results, Discussion, Materials and Methods | Viral engagement with macrophages activates Toll-Like-Receptors ( TLRs ) and viruses must contend with the ensuing inflammatory responses to successfully complete their replication cycle ., To date , known counter-strategies involve the use of viral-encoded proteins that often employ mimicry mechanisms to block or redirect the host response to benefit the virus ., Whether viral regulatory DNA sequences provide an opportunistic strategy by which viral enhancer elements functionally mimic innate immune enhancers is unknown ., Here we find that host innate immune genes and the prototypical viral enhancer of cytomegalovirus ( CMV ) have comparable expression kinetics , and positively respond to common TLR agonists ., In macrophages but not fibroblasts we show that activation of NFκB at immediate-early times of infection is independent of virion-associated protein , M45 ., We find upon virus infection or transfection of viral genomic DNA the TLR-agonist treatment results in significant enhancement of the virus transcription-replication cycle ., In macrophage time-course infection experiments we demonstrate that TLR-agonist stimulation of the viral enhancer and replication cycle is strictly delimited by a temporal gate with a determined half-maximal time for enhancer-activation of 6 h; after which TLR-activation blocks the viral transcription-replication cycle ., By performing a systematic siRNA screen of 149 innate immune regulatory factors we identify not only anticipated anti-viral and pro-viral contributions but also new factors involved in the CMV transcription-replication cycle ., We identify a central convergent NFκB-SP1-RXR-IRF axis downstream of TLR-signalling ., Activation of the RXR component potentiated direct and indirect TLR-induced activation of CMV transcription-replication cycle; whereas chromatin binding experiments using wild-type and enhancer-deletion virus revealed IRF3 and 5 as new pro-viral host transcription factor interactions with the CMV enhancer in macrophages ., In a series of pharmacologic , siRNA and genetic loss-of-function experiments we determined that signalling mediated by the TLR-adaptor protein MyD88 plays a vital role for governing the inflammatory activation of the CMV enhancer in macrophages ., Downstream TLR-regulated transcription factor binding motif disruption for NFκB , AP1 and CREB/ATF in the CMV enhancer demonstrated the requirement of these inflammatory signal-regulated elements in driving viral gene expression and growth in cells as well as in primary infection of neonatal mice ., Thus , this study shows that the prototypical CMV enhancer , in a restricted time-gated manner , co-opts through DNA regulatory mimicry elements , innate-immune transcription factors to drive viral expression and replication in the face of on-going pro-inflammatory antiviral responses in vitro and in vivo and; suggests an unexpected role for inflammation in promoting acute infection and has important future implications for regulating latency . | Here we discover how inflammatory signalling may unintentionally promote infection , as a result of viruses evolving DNA sequences , known as enhancers , which act as a bait to prey on the infected cell transcription factors induced by inflammation ., The major inflammatory transcription factors activated are part of the TLR-signalling pathway ., We find the prototypical viral enhancer of cytomegalovirus can be paradoxically boosted by activation of inflammatory “anti-viral” TLR-signalling independent of viral structural proteins ., This leads to an increase in viral gene expression and replication in cell-culture and upon infection of mice ., We identify an axis of inflammatory transcription factors , acting downstream of TLR-signalling but upstream of interferon inhibition ., Mechanistically , the central TLR-adapter protein MyD88 is shown to play a critical role in promoting viral enhancer activity in the first 6h of infection ., The co-option of TLR-signalling exceeds the usage of NFκB , and we identify IRF3 and 5 as newly found viral-enhancer interacting inflammatory transcription factors ., Taken together this study reveals how virus enhancers , employ a path of least resistance by directly harnessing within a short temporal window , the activation of anti-viral signalling in macrophages to drive viral gene expression and replication to an extent that has not been recognised before . | null | null |
journal.pcbi.1005292 | 2,016 | Inferential Structure Determination of Chromosomes from Single-Cell Hi-C Data | The rapid development of chromosome conformation capture techniques such as 3C 1 , chromosome conformation capture-on-chip 2 , the closely related circular chromosome conformation capture 3 ( both named 4C ) and 5C 4 culminated in a genome-wide variant , Hi-C 5 , which provides all-against-all contact information ., Hi-C experiments confirmed previously established hallmarks of genome organization including the existence of chromosome territories 5 , 6 and led to important new findings such as the partitioning of chromosomes into alternating active and passive , megabase-sized compartments 5 , 7 and in non-tissue-specific topologically associating domains ( TADs ) on the sub-megabase scale 8–10 ., Chromosome conformation capture experiments typically analyze populations of millions of cells , thereby only providing a population-averaged view ., Recently , however , Nagano et al . 11 pioneered Hi-C on single cells by executing all of the steps of the original Hi-C protocol within permeabilized cells and selecting individual cells for further analysis ., Although the single-cell Hi-C approach provided only very sparse contact data , the structural information was sufficient to reveal unprecedented insights into genome organization including a high cell-to-cell variability of interdomain and trans-chromosomal contacts as well as the persistence of TADs across single cells ., Many structural insights such as the existence of TADs or the scaling behavior of contact probabilities with genomic distance can be found by analyzing genome-wide contact matrices ., Nevertheless , it seems attractive to obtain a more direct view of the 3D architecture of genomes by structural modeling based on the experimental contact information ., To compute representative 3D structures of genomes , various approaches have been explored ., There is a growing array of computational methods for calculating consensus structures from population Hi-C data ., Typically , these methods first derive distances from the experimental contact frequencies by using different heuristics ., In the early work by Duan et al . 12 , a model of the yeast genome was computed based on data from a 4C-related experiment ., Bau et al . 13 mapped inverse log Z-scores from 5C data to distances and used the Integrated Modeling Platform ( IMP ) 14 to compute structural models ., PASTIS 15 addresses chromosome structure determination by means of maximum likelihood , whereas ChromSDE 16 relies on semi-definite programming ., Along with the structure , both PASTIS and ChromSDE optimize an additional free parameter , which is used to translate contact counts into distances ., Trieu et al . 17 used an optimization-based approach , but modeled contact counts explicitly ., Also Zhang et al . 18 avoided the conversion of contact counts to distances , but instead of a consenus structure , they obtain structure ensembles by simulating from an approximate energy landscape for chromosomes ., Two major challenges complicate the adaptation of methods for chromsome structure inference from population Hi-C to single-cell data ., First , single-cell Hi-C measures only the formation of a contact rather than contact frequencies ., Second , only a small subset of all chromosomal contacts is measured such that the contact information is very sparse ., Therefore , specialized methods for the analysis of single-cell Hi-C contacts need to be developed ., Multidimensional scaling ( MDS ) is a popular method to obtain three-dimensional structures from incomplete and noisy distance information and was already used in the first publication on chromosome conformation capture 1 ., A major limitation of MDS is that with dwindling number of data serious artifacts are introduced , which eventually leads to a complete break-down of the procedure ., Shortest-path reconstruction in 3D ( ShRec3D ) 19 and an approach employing manifold-based optimization ( MBO ) 20 are two recent variants of MDS that aim to overcome these challenges for single-cell Hi-C data ., Both methods define a contact distance between loci that show a contact in the Hi-C experiment and introduce a similar distance between neighboring loci along the chromatin fiber ., The missing entries of the distance matrix are imputed by shortest-path distances , which are computed for a graph derived from the experimental contacts and the fiber connectivity ., ShRec3D applies MDS directly to the completed distance matrix , whereas Paulsen et al . downweigh the shortest-path distances and utilize optimization techniques on matrix manifolds 21 ., Nagano et al . 11 used restraint-based modeling to obtain structures of the X chromosome from their single-cell data ., They derived distances from the contact data and combined the restraint energy with a simple polymer model ., To find chromosome structures that fit the restraints , they used simulated annealing ( SA ) combined with molecular dynamics ., However , the application of optimization approaches such as SA or MDS to chromosome structure determination suffers from the same conceptual problems described by Rieping et al . 22 in the context of protein structure calculations from NMR data ., First , the scoring function typically involves model parameters that are unknown and set to ad hoc values ., An example is provided by the weighting factors that define the “strictness” of the restraints 23 ., Second , structure ensembles generated by minimization approaches lack a sound statistical foundation ., These ensembles are computed by running multiple minimizations from randomly varying initial structures ., Although this practice seems plausible , the variability of the ensemble is not a valid “structural error bar” , because it does not only reflect the quality and amount of the data , but also the power of the optimization procedure , to mention but one reason ., Third , minimization approaches fail to clearly separate model parameters from algorithmic parameters , again blurring the meaning of the structure ensemble ., To address these issues , Rieping et al . 22 introduced Inferential Structure Determination ( ISD ) as an unbiased and parameter-free alternative to minimization approaches to biomolecular structure determination ., ISD is a Bayesian probabilistic framework that views biomolecular structure determination as an inference problem ., At the core of the ISD approach is a probability distribution over conformational space representing and combining both noisy and possibly incomplete data as well as prior knowledge about the unknown structure ., Several Bayesian approaches have been developed to model the structure of chromosomes based on ensemble Hi-C data 24 , 25 ., These methods try to infer a consensus structure of the population data , which is only of limited use ., Moreover these approaches have not fully benefited from the Bayesian approach to chromosome structure inference ., Wang et al . 26 calculate structural ensembles using a Bayesian approach , but only optimize the posterior probability and do not apply the full Bayesian inference machinery , which allows for the quantification of parameter uncertainties and the comparison of alternative models ., Here we report on the application of ISD to infer statistically well-defined ensembles of chromosome structures from single-cell Hi-C data ., We show that Markov chain Monte Carlo ( MCMC ) sampling allows us to compute diverse ensembles of coarse-grained chromosome conformations that reflect the sparsity of single-cell Hi-C contacts ., MCMC techniques and the flexibility of our Bayesian approach also allow us to compare different models of the chromatin fiber as well as alternative models for Hi-C contacts ., We use the conformational ensembles to map epigenetic marks into three-dimensional space ., Furthermore , we demonstrate that ISD outperforms alternative methods on simulated data ., Finally , we show how to extend the approach to diploid chromosomes and infer the structures of two chromosome copies simultaneously ., We model the chromatin fiber with a beads-on-a-string representation ., Owing to the sparsity of single-cell Hi-C contacts , we use a highly coarse-grained model in which every bead represents 500 kb of chromatin and has a radius of approximately 215 nm ., Beads are connected such that they form a linear chain ., The connectivity is enforced by a harmonic backbone potential , which penalizes distances between consecutive beads as soon as they exceed the bead diameter a ., Beads are soft and allowed to overlap to some extent ., We investigated two volume exclusion terms: a purely repulsive potential under which two beads that are closer than their diameter repel each other , and a Lennard-Jones potential with repulsive and attractive contributions ., The sum of the backbone and nonbonded potential are part of the prior distribution ( see Materials and Methods for details ) ., We first studied the properties of our model for the single-copy X chromosome of male mouse , which measures 166 Mb in length , and which we represent using 333 beads ., We generated structures from the prior distribution and reconstructed the distribution of the radius of gyration Rg as a measure for the compactness of the fiber ., Because there are no or only weak attractive interactions between the beads , the vast majority of structures generated from the prior showed an extended conformation ., Fig 1A displays the density of states ( or equivalently the microcanonical entropy ) as a function of the radius of gyration ., The density of states counts how many chromosome conformations map to a particular Rg value ., There is a strong entropic force that pushes the fiber into an extended state characterized by a large radius of gyration ., For the repulsive excluded volume term we found Rg = 11 . 2 ± 1 . 1 μm; for the Lennard-Jones potential we have Rg = 8 . 6 ± 1 . 3 μm ., Due to the attractive contribution in the excluded volume term , the Lennard-Jones potential shows a higher preference for compact structures ., Yet for both potentials , the fraction of compact structures with Rg smaller than 2 μm , say , is vanishingly small: 1 . 8 × 10−24 for the quartic repulsion potential and 7 . 8 × 10−21 for the Lennard-Jones term ., The preference of the prior probability for extended structures with large average radii of gyration is incompatible with fact that chromosomes localize in chromosome territories , i . e . relatively compact subcompartments of the nucleus 6 ., To good approximation , the size of the chromosome is a function of the radius of gyration ., Empirically , we found that 1 ., 21 × R g 1 ., 11 gives a good estimate of the size of the X chromosome for relatively compact structures ( Fig 1B ) ; a simpler , yet precise enough relation is 2 . 58 × Rg ., Using this approximation , the experimental chromosome size measurement of 3 . 7 ± 0 . 3 μm obtained with X-chromosome paint FISH 11 corresponds roughly to Rg ≈ 1 . 43 ± 0 . 12 μm ., Therefore , the average radii of gyration reported above are an order of magnitude too large compared to the experimental finding ., To incorporate the information from FISH into our probabilistic chromosome model and thereby inform the prior probability about the expected chromosome size , we assume a Gaussian error model for the chromosome size measurements ., Based on our approximate relation between chromosome size and Rg , this term corresponds to a harmonic radius of gyration restraint with an experimental Rg value of 1 . 43 ± 0 . 12 μm ., By using probability calculus , it is possible to combine single-cell Hi-C contacts with the FISH data and our chromosome models ( see Materials and Methods ) ., Nagano et al . 11 analyzed Th1 cells of male mouse and selected ten cells that passed various quality criteria ., We first focused on the most promising data set from cell 1 showing 616 X-chromosomal cis-contacts , which represents the highest number of contacts among all ten cells ., We mapped these contacts onto the 500 kb beads; the removal of intra-bead contacts ( “self-contacts” ) resulted in 438 contacts , out of which 399 were unique ( some contacts mapped onto identical pairs of beads ) ., We used a logistic model to quantify the probability that a cis-chromosomal contact is observed in a Hi-C experiment ., The logistic function is a smooth version of a step function and has probability close to one if the contact is formed in the model structure of the fiber , and vanishes if the beads are too far apart ., Consistent with other approaches 27 , 28 , we chose a distance cutoff dc = 1 . 5 × a ≈ 650 nm to decide whether two beads i and j are in contact ., The smoothness of the logistic function was chosen such that a distance violation of 2 . 5% of the contact distance dc has a probability smaller than 10−6 ., We generated ensembles of X-chromosome structures for both excluded volume potentials , both without and with the additional model for FISH data introduced in the previous section ., To sample chromosome conformations we used Hamiltonian Monte Carlo ( HMC ) 29 , a stochastic variant of molecular dynamics ., We started the HMC simulation from a fully extended X-chromosome structure ., To ensure correct conformational sampling , we used replica exchange Monte Carlo 30 , 31 which runs multiple HMC simulations at different temperatures in parallel and exchanges conformations between the different simulations ( see Materials and Methods ) ., Replica exchange simulations are among the most powerful Monte Carlo methods to simulate complex probability distributions but computationally demanding ., It is also possible to generate X-chromosome conformations that satisfy the experimental contacts by only running HMC at a fixed temperature ., The HMC sampler rapidly finds a compact structure that fits the contact data very well without producing significant violations ., Nevertheless , the following results were obtained by running replica exchange simulations ., As a first validation of our inference approach we studied whether all experimental contacts can be satisfied in the 3D model of the X chromosome ., Fig 2A shows that this is indeed the case ., For all four prior distributions the number of violations fluctuates about a small percentage of less than 3% ( see also S1 Fig ) ., By increasing the steepness of the logistic function , we could decrease this number to exactly zero ., As a further validation we analyzed the average pairwise distance matrix computed from the sampled X-chromosome structures ., Fig 2B and 2C compare the experimental contacts with the average distance matrix ., We observe that loci with a high number of cis-chromosomal contacts correspond to patches of small pairwise distances in the distance matrices ., Using a diagonal permutation test , we found that the average distance matrices based on the different prior probabilities agree to a large extent , which underlines the fact that the prior does not have a strong influence on the average properties of the structure ensemble ., The correlation of the distance matrices ranges between 93% and 95% ( see S1 Fig for details ) ., Nevertheless , we can use Bayesian model comparison to study whether the data show any preference for one of the prior probabilities ., To do so , we estimated the model evidence ( also known as marginal likelihood ) from the MCMC simulations ., The model evidence quantifies how likely a probabilistic model is in the light of the data ., Fig 2D shows that the contact data prefer the Lennard-Jones term over the quartic repulsion term ., The incorporation of the information from FISH raises the model evidence further ( see S1 Table ) ., Fig 2E shows the radii of gyration obtained without and with FISH data when using the Lennard-Jones potential ., The structures generated with the cis-chromosomal contacts only , without additional FISH data are already quite compact with an average Rg of 1 . 32 μm ., Due to the strong forces exerted by the logistic contact restraints , the ensemble is slightly more compact than suggested by the FISH data ., When incorporating the FISH term , the average Rg shifts towards larger values with an average of 1 . 38 μm , corresponding to a chromosome size of ∼3 . 6 μm ., The FISH data do not compromise the fit with the contact restraints: the number of violations does not change upon incorporation of the Rg model ( see S1 Fig ) ., However , because the additional radius of gyration term helps to focus the conformational sampling on reasonably compact chromosome structures , the model evidence of the FISH based posterior is higher than without FISH data ., By looking at the variance of the pairwise distances in the structure ensembles ( Fig 2F and S1 Fig ) , we find that the regions at the start of the X chromosome and around the centromere show the highest degree of conformational diversity ., It is unclear , however , if this increased variability reflects true conformational fluctuations or simply the fact that these regions are unmappable ., Nonetheless , we also observe an overall rise in the distance fluctuations towards the telomeric region , which might indicate that this part is indeed more dynamic ., Many approaches to infer chromosome conformation from Hi-C data resort to modeling based on distance restraints ., To this end , pairwise distances need to be derived from the experimental contact information ., For example , the contact frequencies measured in ensemble Hi-C experiments were converted to distances by assuming a power law that relates the contact probability to the inter-bead distance , which is motivated by results from polymer physics ( see e . g . 15 ) ., In case of single-cell data , this is more challenging because only single contacts are observed and not contact frequencies ., To interpret the observation of a single-cell Hi-C contact as a distance measurement , we introduce an unknown distance δ between two loci that will be crosslinked ., For the unknown chromosome conformation X it should therefore hold that δ ≈ dij ( X ) where dij ( X ) is the model distance between beads i and j representing both loci ., Because δ is unknown , we estimate this model parameter simultaneously with the chromosome structure ., In contrast , Nagano et al . 11 used a / n i j 2 as experimental distance where a denotes the bead diameter and nij counts how often beads i and j form a contact after mapping the high-resolution contacts onto the coarse-grained representation of the fiber ., At 500 kb resolution nij ranges from 1 to 3 ., In our approach , the repeated occurrence of a contact ( nij > 1 ) does not lead to a shortening of the contact distance , but rather to an enforcement of the distance restraint , which is duplicated nij times ., Due to experimental errors and shortcomings of our model , we have to account for discrepancies between the unknown experimental distance and the model distances ., This is achieved by introducing a probabilistic model for the distribution of the discrepancy between δ and dij ( X ) ., We studied two error distributions: The first assumes a Gaussian shape with a flat plateau for distances between δ ± 0 . 2 × a in accordance with the approach by Nagano et al . 11 ., The second model is a lognormal distribution 32 ., Both error models depend on an additional unknown error parameter σ , which reflects how well the experimental distance agrees with the model distance ., The inverse variance w = 1/σ2 can be interpreted as the weight of the distance restraint potential 23 ., We set w to relatively large values to reflect our assumption that all observed contacts are correct ( which we also assumed in the logistic contact model ) ., We used w = 100 for the Gaussian with flat plateau and w = 500 for the lognormal model , because it has a softer shape than the harmonic restraint resulting from the Gaussian model ., To estimate the experimental distance using ISD , we rewrite δ = a/γ where γ > 0 is an unknown scaling parameter such that ideally a = γdij ( X ) ., Fig 3A shows histograms of the estimated distance scale γ ., The distance scale γ attains similar values for both error models ., The average values are 0 . 724 ± 0 . 003 for the Gaussian with a flat plateau and γ = 0 . 744 ± 0 . 018 for the lognormal model ., These values translate into an estimated inter-locus distance of ∼1 . 4 × a , which is comparable to the contact distance dc assumed in the logistic contact model ., The distances involved in an experimental contact ( shown in Fig 2A for the contact model ) are less restrained in the distance-based models ( see S2 Fig ) ., Nevertheless the agreement between the structure ensembles generated with both distance-based models is fairly high ., The correlation between the average distance matrices is 95% ( Fig 3B ) , and both ensembles also agree well with the ensemble based on the logistic contact model ( with a correlation coefficient of 96% between the average distances matrices generated by each of the distance-based models and the contact model ) ., Using Bayesian model comparison , we can also answer which of the two error models , Gaussian with a flat plateau or lognormal model , is preferred by the experimental data ., Fig 3C shows the model evidence Pr ( D|I ) as a function of the replica temperature parameter λ ., The values for λ = 1 indicate that the data tell us to prefer the Gaussian model with a flat plateau over the lognormal error model ., By running calculations in which we varied the error parameter σ , we found that the exact value of the distance scale γ depends on the assumption that we make about the reliability of the Hi-C data to some extent ( see S3 Fig ) ., However , for reasonably small σ/large w this dependence is less strong and the distance scale γ reaches a plateau ., Bayesian inference also allows us to estimate both model parameters , γ and σ , simultaneously , but this results in rather broad ensembles that do not adopt a well-defined structure ( see S1 Appendix ) ., The failure to generate well-defined structure ensembles , if w is allowed to vary is due to the sparsity of the data: The intrinsic tendency of the chromatin fiber to adopt a disordered state is stronger than the forces exert by the distance restraints with variable weight ., Only with a large enough weight it is possible to counter-balance the entropic forces ., Another remedy is to improve the model of the chromatin fiber 33 , but this is beyond the scope of this article ., Our results show that it is possible , in principle , to model single-cell Hi-C contacts as distance measurements ., However , we will use the logistic contact model in the remainder of this article , because single-cell Hi-C observes binary contacts rather than continuous distances ., We now take a closer look at the ensembles generated with the ISD approach and compare them to the published ensemble by Nagano et al . 11 ., The structure of the X chromosome adopts a bipartite conformation formed by two super-domains that approximately span the centromeric half ( ∼1–100 Mb ) and the telomeric half ( ∼100–166 Mb ) of the chromatin fiber ., This large-scale domain structure is readily apparent from both the experimental contacts and the average distance matrix ( Fig 2C ) and becomes immediately visible in an explicit representation of the structure ensemble ( Fig 4A ) ., Cluster analysis reveals that the ISD ensemble comprises multiple principal conformations about which the structures fluctuate ., Closer inspection shows that the cluster centers are partial mirror images of each other ., None of the likelihood and prior factors contributing to the posterior distribution distinguishes between a particular bead configuration and its mirror image , because all factors depend on distances only ., Moreover , since there are only few contacts between the super-domains , each super-domain can show two conformations which are mirror images of each other ., This results in at least four possible chromosome conformations which all achieve a similar goodness of fit of the Hi-C contacts ., Our cluster analysis finds that the eight most dominating structural clusters produced by ISD cover ∼90% of all states sampled from the posterior ( see S4 Fig for further details ) ., The four most highly populated clusters are shown in Fig 4A ., These are approximate mirror images of each other ., We applied the same type of cluster analysis also to an ensemble of 200 X-chromosome structures computed by Nagano et al . 11 ( see S5 Fig ) ., We found similar structural clusters that are partial mirror images of each other and have corresponding structural clusters in our ensemble ( details given in Supplementary Information ) ., However , overall the ISD ensemble seems to be more diverse , showing more clusters than the ensemble by Nagano et al . Visual inspection of the structural clusters suggests that the overall variability in the ISD ensemble is quite high and comparable to the fluctuations in the ensemble by Nagano et al . But each cluster of the ISD ensemble appears to be slightly better defined than the clusters in the ensemble by Nagano et al . This might be due to the more exhaustive sampling achieved by our Monte Carlo algorithm , the attractive contributions in the excluded volume term and the fact that we model Hi-C measurements as contacts with a steep sigmoidal contact probability rather than distance restraints ., For each cluster , we studied the local variability of the beads by using standard techniques for the analysis of NMR structure ensembles ., We estimated the local precision of the bead positions by the root mean square fluctuation ( RMSF ) after superposition of the cluster members onto the cluster center ., Fig 4B shows RMSF curves for both ensembles ., The RMSF curves from the four major clusters of the ISD ensemble correlate almost perfectly , the same is true for the clusters of the previously published X chromosome ensemble ., We also find a high agreement between the RMSF fluctuations in the ISD clusters and the fluctuations within the clusters of the ensemble by Nagano et al . The average Pearson correlation coefficient of the RMSF profiles is 91% , showing that the ensembles obtained with both approaches agree strongly in their conformational heterogeneity ., Together with the high correlation of 85% between the average distance matrices of the ISD ensemble and the ensemble by Nagano et al . ( see S2 Table ) , these findings indicate that both ensembles show many similar properties ., A tube representation of the local variability of the four principal conformers is shown in panel 4A ., Again , we find a higher conformational diversity towards the telemore , which was already apparent from the standard deviation of pairwise bead distances ( Fig 2F ) ., We also ran ISD simulations on contact data from 5 additional Th1 cells ., The average distance matrices indicate that the ensembles are significantly different , indicating the cell-to-cell variability of chromosome conformations found by Nagano et al . 11 ( see S6 and S7 Figs ) ., A comparison of the ensembles obtained with data from cell 1 to cell 6 is reported in S2 Table ., This comparison shows that there is an overall agreement between the X-chromosome conformations obtained with the ISD approach and the restraint-based modeling approach by Nagano et al . also for the data sets from other cells ., By comparing the average distance matrices ( S6 Fig ) , we conclude that chromosome structures from different cells share some common features , such as the partitioning into more or less well-defined domains , which appear as blocks of small distances along the diagonal ., The size and location of these domains can differ significantly from cell to cell ., While in cells 1 , 4 , 5 and 6 , a well-defined telomeric domain is visible as a block between ∼130 − 166 Mb , it is much less pronounced in cells 2 and 3 ., This difference is also evident in the structural models ( S7 Fig ) , in which the telomeric domain appears as a separated structural domain ( colored in red ) ., In the models for cell 4 , the telomeric domain is on average ∼10 Mb shorter than in the models for cell 1 , 5 and 6 ., The six different models also exhibit considerable differences in the spatial proximity of loci that are far apart in sequence ., An example involves the centromeric region on the lower left of the average distance matrices ., In the ensemble representing the X chromosome in cell 6 , this region is spatially close to many loci that are up to ∼100 Mb distant in sequence , while in cell 5 , this region shows significantly larger distances to most other loci ., S7 Fig confirms this by showing that the blue and cyan parts of the X chromosome structure are much more exposed in cell 5 than in cell 6 ., Taken together , a picture of highly variable , presumably stochastic chromosome organization emerges , in which conformations nevertheless share large-scale properties across different cells ., A problem with current Hi-C based chromosome modeling is that it is difficult to validate the calculated structures ., However , there are some independent sources of information , not used during modeling , that should be consistent with a meaningful structure ensemble ., One is the information provided by population Hi-C ., Although population Hi-C looks at a large pool of cells , the information about the absence of contacts should also hold for the chromosome structure based on single-cell data ., As can be seen from S8 Fig by comparing the contact probabilities derived from the ISD ensemble with population Hi-C data , the ISD ensemble based on single-cell data indeed avoids contacts between loci that have a low contact probability in the population Hi-C map ., Another validation is provided by the location of beads involved in trans-chromosomal contacts ., Fig 4C and S9A Fig show that beads which are engaged in contacts with loci on other chromosomes tend to accumulate on the periphery of the structure ensemble , which is indirect evidence in support of our chromosome ensembles ., Based on the inferred structure ensemble it becomes possible to generate three–dimensional maps of genomic and epigenetic features and to correlate the features spatially ., Fig 4C shows a volume representation of chromosomal regions that are enriched in H3K4me3 and lamin-B1 associated domains in the first structural cluster ., Loci that are enriched in these epigenetic marks tend to aggregate in three-dimensional space ., The lamin associated domains as well as H3K4me3-enriched regions both show a tendency to locate in the periphery of the X chromosome , where they occupy distinct regions ., In accord with previous findings by Nagano et al . , we also find that H3K4me3 is enriched in some parts of the interior of the X chromosome ( see S9C Fig ) ., Due to the sparsity of the chromosomal contacts , we have used a very coarse-grained representation of the chromatin fiber ., At this resolution , we can only study large-scale chromosomal organization ., Higher resolution representations are typically needed to gain biologically relevant insights into 3D chromosome organization ., We therefore also applied the ISD approach using a ten-fold higher resolved chromatin fiber ., Each bead now represents 50 kb of chromatin , thereby matching the finest resolution used by Nagano et al . 11 ., At this resolution the radius of a bead amounts to ∼100 nm ., At 50 kb resolution , we represent the X chromosome with 3330 spherical beads ., We generated structure ensembles based on the Lennard-Jones volume exclusion term and the additional FISH restraint ., We modeled the intra-chromosomal contacts with the logistic model; in contrast to Nagano et al . no additional “anti-contact” restraints from the ensemble Hi-C matrix were introduced ., To compare the overall properties of the structure ensembles generated at 500 kb and 50 kb resolution , we downsampled the distance matrices from the 50 kb models to match the resolution of the coarse-grained models ., Downsampling was achieved by averaging 10 × 10 patc | Introduction, Results, Discussion, Materials and Methods | Chromosome conformation capture ( 3C ) techniques have revealed many fascinating insights into the spatial organization of genomes ., 3C methods typically provide information about chromosomal contacts in a large population of cells , which makes it difficult to draw conclusions about the three-dimensional organization of genomes in individual cells ., Recently it became possible to study single cells with Hi-C , a genome-wide 3C variant , demonstrating a high cell-to-cell variability of genome organization ., In principle , restraint-based modeling should allow us to infer the 3D structure of chromosomes from single-cell contact data , but suffers from the sparsity and low resolution of chromosomal contacts ., To address these challenges , we adapt the Bayesian Inferential Structure Determination ( ISD ) framework , originally developed for NMR structure determination of proteins , to infer statistical ensembles of chromosome structures from single-cell data ., Using ISD , we are able to compute structural error bars and estimate model parameters , thereby eliminating potential bias imposed by ad hoc parameter choices ., We apply and compare different models for representing the chromatin fiber and for incorporating singe-cell contact information ., Finally , we extend our approach to the analysis of diploid chromosome data . | Spatial interactions between distant genomic regions are of fundamental importance in gene regulation and other nuclear processes ., Recent chromatin crosslinking ( “Hi-C” ) experiments probe the spatial organization of chromosomes on a genome-wide scale to an extent that was previously unattainable ., These experiments report on contacting loci and thus provide information about the three-dimensional structure of the genome ., Unfortunately , the data are noisy and do not determine the structure uniquely ., There is also little quantitative prior knowledge about the large-scale organization of chromosomes ., Here , we address these challenges by developing a Bayesian statistical approach that combines a minimalist polymer model with chromosome size measurements and conformation capture data ., Our method generates statistical ensembles of chromosome structures from extremely sparse single-cell Hi-C data ., We remove potential bias by learning modeling parameters from the experimental data and apply model comparison techniques to investigate which among a set of alternative models is most supported by the Hi-C data ., Our method also allows for modeling with ambiguous contact data obtained on polyploid chromosomes , which is an important step towards three-dimensional modeling of whole genomes . | chromosome structure and function, mathematics, statistics (mathematics), protein structure, epigenetics, structural genomics, chromatin, research and analysis methods, sex chromosomes, chromosome biology, proteins, mathematical and statistical techniques, gene expression, x chromosomes, monte carlo method, molecular biology, genetic loci, biochemistry, chromosomes, cell biology, genetics, biology and life sciences, physical sciences, genomics, statistical methods, macromolecular structure analysis | null |
journal.pcbi.1004106 | 2,015 | Spatio-temporal Model of Endogenous ROS and Raft-Dependent WNT/Beta-Catenin Signaling Driving Cell Fate Commitment in Human Neural Progenitor Cells | Canonical WNT signaling is a central pathway in embryonic development and adult homeostasis , while its aberrant form is involved in a number of human cancers and developmental disorders 1–3 ., The WNT/β-catenin signal transduction is characterized by a reaction cascade , that is initiated by extracellular WNT molecules and eventually leads to an accumulation of cytosolic β-catenin and its subsequent shuttling into the nucleus ., In the nucleus β-catenin associates with the Lef/Tcf transcription factors triggering a pathway-specific gene response relevant for the regulation of various physiological and developmental processes , including neuronal differentiation 3 , 4 Accordingly WNT/β-catenin signaling has been reported to be involved in the neuronal differentiation process of human neural progenitors cells ( hNPCs ) 5 ., NPCs provide a new , promising basis for the in-vitro growth of neuron populations that can be used in replacement therapies for neurodegenerative diseases , such as Parkinson’s or Huntington’s diseases 6 , 7 ., However , controlling NPC differentiation in stem cell engineering demands a thorough understanding of neuronal and glial cell fate determination and its endogenous regulation ., A first characterization of ReNcell VM197 hNPC cell fate commitment uncovered a spatio-temporal regulation of WNT/β-catenin key proteins , like LRP6 , DVL , AXIN and β-catenin throughout the entire phase of early differentiation 8 ., However , the exact mechanisms that drive the WNT/β-catenin signaling and therewith control the cell fate commitment in hNPC remain unclear ., One of the key mechanisms of the WNT signal transduction is the formation of a large protein-receptor complex , called signalosome , in response to the extracellular WNT stimulus 9 , 10 ., The signalosome consists of the membrane-integral receptors FZ and LRP6 and several cytosolic proteins , like Dishevelled ( DVL ) , CK1γ , AXIN and GSK-3β ., The stable aggregation of the signalosome triggers the phosphorylation of several intracellular phosphorylation sites ( mainly PPSPXS motifs ) in the cytosolic tail of LRP6 , generating high-density platforms for the recruitment of AXIN 11–13 ., Due to the binding of AXIN ( and GSK-3β ) to LRP6 , key components of the destruction complex are inhibited , which in turn leads to an accumulation and translocation of β-catenin into the nucleus and eventually to the well known gene transcription signal ., Recently , several studies suggested an involvement of lipid rafts in the WNT/β-catenin pathway 14–17 ., Lipid rafts are local assemblies of highly concentrated sphingolipids and cholesterol in the cell membrane 18 ., The diffusion inside rafts is significantly slowed down , which in turn influences the general diffusion and localization of transmembrane receptors 19 , 20 ., Apparently , the localization of LRP6 in lipid rafts is crucial for its successful phosphorylation , implying a major impact of lipid rafts on the activation of signalosome , hence WNT/β-catenin signaling 15 , 17 ., Therefore we investigate the mutual influence of lipid rafts on WNT-signaling during the in-vitro differentiation of immortalized human neural progenitor cells ( ReNcell VM197 ) ., The ReNcell VM197 cell line was derived from the ventral mesencephalon region of a human fetal brain tissue and is characterized by a rapid differentiation ., Upon growth factor removal ReNcell VM197 cells differentiate into neurons and glial cells within a few days and without any additional external stimulation ., This allows us to study WNT signaling in the context of cell fate commitment in a time dependent manner ., During our investigations we found that lipid raft disruption by Methyl-β-Cyclodextrin ( MbCD ) effectively inhibits WNT/β-catenin signal transduction ., This implies that raft disruption serves as effective inhibitor for WNT/β-catenin signaling in our cell line ., However , surprisingly we found that immediately after the initiation of differentiation , raft-deficient cells still show a transient β-catenin signaling activity , raising the question what triggers the early immediate response despite the apparent WNT/β-catenin signaling inhibition ?, In a recent study we showed that during the initiation phase of ReNcell VM197 differentiation an early spontaneous production of reactive oxygen species ( ROS ) occurs , which promotes a DVL-mediated downstream activation of canonical WNT signaling 21 ., ROS are chemically reactive radical and non-radical molecules containing molecular oxygen mainly generated as by-products of the electron transfer pathway in the mitochondrial respiratory chain 22 ., Excessive ROS accumulation induces cell damage through an oxidative stress involved in various pathologies as diabetes , cardiovascular diseases or neurological disorders ., However , if present in moderate amounts , ROS have been implicated as signaling mediators in various physiological processes i . e . activation of Rac1 , PI3K , MAPK cascade , ASK1-dependent apoptosis , p21-mediated signaling , or modulation of thioredoxin-dependent transcription factors 23 , 24 ., A few recent studies found an involvement of ROS in the regulation of canonical WNT signaling while direct proof that ROS metabolism acts as endogenous transmitters were missing since ROS implication has been reported through the use of exogenous stimulation by pro-oxidant compounds 25 or injury 26 ., However further evidence was provided in our previous study on ReNcell VM197 cells , as an early increase of mitochondrial ROS metabolism after growth factor removal was found to modulate DVL-mediated WNT/β-catenin pathway and neurogenesis 21 ., To evaluate , whether an interplay between ROS-induced and lipid raft dependent WNT/β-catenin signaling can explain our experimental results we apply computational modeling ., We extend the current standard model of the WNT/β-catenin pathway 27 with the aforementioned membrane-related processes including lipid rafts/receptor dynamics and combine this with an intracellular ROS/β-catenin signaling mechanism ., The model is based on experimental data as well as literature values and has been extensively validated against in-vitro and in-silico data under a wide range of varying conditions ., In the following we describe experimental data , retrieved from ReNcell VM197 human progenitor cells ., The ReNcell VM197 is a well-characterized cell line , that has been successfully applied in several studies and proven to be a simple and accepted model to investigate different aspects of neural differentiation 5 , 8 , 30–32 ., The major advantage of this cell line is its rapid differentiation ., Within three days after growth factor removal , ReNcell VM197 cells differentiate into neurons , astrocytes , and oligodendrocytes without any additional exogenous stimulation ., We evaluate the impact of lipid raft disruption on WNT/β-catenin signaling during differentiation by measuring the temporal progress of WNT signaling in terms of nuclear β-catenin concentrations in methyl-β-cyclodextrin-treated and untreated cells in the process of cell fate commitment ., Accordingly proliferating ReNcell VM197 cells were used as reference ( 0h ) , whereas all following time points were measured after initiating the differentiation by growth factor removal ., Note , that we only consider the first 12 hours after induction of differentiation ., Typically most of the cells commit themselves for differentiation within the first 12 hours ., Also , at later time points the cell population of ReNcell VM197 is already so heterogeneous due to differentiation , that potential signal activities may originate from multiple sources ., Fig . 2 shows a schematic representation of our basic WNT model , i . e . the two main model components of membrane-related LRP6/CK1γ and axin/β-catenin signaling and their interaction ., The model is defined in ML-Rules , a hierarchical , multi-level modeling language 34 ., The model is stochastic , multi-compartmental and completely based on mass action kinetics ., For a more detailed introduction of ML-Rules and for the implementation of the basic WNT/β-catenin model see Supporting Information ( S1 Text and S3 Text ) ., We demonstrated , that the combined membrane and axin/β-catenin model captures relevant processes of canonical WNT signaling and is able to predict the WNT/β-catenin dynamics in response to arbitrary WNT stimuli of untreated cells with undisturbed lipid rafts ., Though , the model is not capable of reproducing the transient activation in raft-deficient cells ( see Fig . 4B ) ., To predict this apparently WNT-independent signal , the present WNT/β-catenin model has to be extended by a presumingly intracellular mechanism ., In a recent study with the same cell line , we uncovered an endogenous , WNT-independent activation of WNT/β-catenin signaling through reactive oxygen species ( ROS ) in response to initiation of differentiation through growth-factor removal 21 ., Thereby an increase of the intracellular ROS level releases the redox-sensitive binding between NRX and DVL , hence promoting a DVL-mediated stimulation of the downstream WNT/β-catenin signal transduction , which eventually leads to the well known β-catenin accumulation in the nucleus ., In fact , several experimental studies demonstrating a redox-dependent activation of WNT/β-catenin signaling have emerged recently ., Funato et . al . reported a robust activation in response to exogenous ROS stimulation in proliferating cells 25 , while Love et . al . showed that injury-induced ROS is required to activate WNT/β-catenin pathway in the context of cell regeneration 26 ., Whereas extensive ROS stimulation may cause oxidative stress and cell damage , it is meanwhile well accepted , that ROS can also act as intracellular messenger inducing redox-sensitive signal transductions when present at physiological concentrations 54 ., To evaluate , whether an interplay between redox- and lipid raft dependent , autocrine WNT/β-catenin activation is a suitable hypothesis to explain our data , we extend our model with a redox-dependent/β-catenin pathway ., Since quantitative experimental data is rarely available , we base our model upon the findings of Funato et . al . and our own recent experimental results 21 ., As depicted in Fig . 5 we extend the given model by a new , redox-dependent model component ., For a complete implementation of the extended model see Supporting Information ( S9 ) ., According to the aforementioned studies , ROS molecules release the redox-sensitive binding of DVL and Nucleoredoxin ( NRX ) , leading to a spontaneous increase in cytosolic DVL concentration ( R36 ) ., Due to the property of DVL to self-associate in a reversible and concentration-dependent manner ( R31–R33 ) , DVL forms self-assemblies that serve as dynamic recruitment platform for AXIN 55 , 56 ( R37/R38 ) ., DVL-bound AXIN is not available for the destruction complex , hence β-catenin can accumulate and translocate into the nucleus ., The spontaneous release of DVL through ROS obviously mimics an overexpression of DVL , which has been demonstrated to trigger WNT/β-catenin signaling , bypassing the requirement for WNT ligands 25 , 55 , 57 ., After successful calibration we connect the ROS/β-catenin model with the model presented in the previous section ., To initiate ROS/β-catenin signaling , we introduce a transient ROS signal at the beginning of differentiation 21 ., This corresponds to the significant increase of endogenous ROS levels measured in ReNcell VM197 human progenitor cells ., Indeed , the extended model is now able to reproduce the immediate β-catenin activation in raft deficient cells as well as the kinetics in untreated cells ( cf . Fig . 6 ) ., Please note that , all parameter values , but the coverage of rafts ( 25% and 0 ) , are exactly the same for the simulation of control and raft-deficient cells ., For both simulation experiments the model is parameterized with a transient ROS signal and a delayed , constant WNT production ., Thus in contrast to the previous model configuration we replaced the initial amount of WNT molecules with a onetime release of ROS molecules ( nRos = 10000 ) in response to growth factor removal ., All other remaining parameter values of our earlier model remain the same , in particular , the delayed ( 90min ) and constant WNT production ( kWsyn = 1 . 9 ) ., The necessity to include such a delay can be explained by inspecting our results more closely: Note , that the increase of β-catenin concentration during the immediate early response ( 1h ) is not significantly different between control and raft deficient cells ., If WNT signaling was directly activated after induction of differentiation , the signal at 1 hours would add up with the β-catenin activation induced by ROS , hence most likely be significantly higher in control than in raft deficient cells ., As this is not the case , we conclude , that the described autocrine , raft-dependent WNT signaling can only be initiated after a certain delay ., However , this also implies that the signal after one hour is entirely based upon WNT/LRP6 independent mechanisms like the presented redox-dependent DVL/β-catenin pathway ., This becomes even more evident , when considering the localization and binding state of AXIN during signaling ( cf . Fig . 6C ) ., While unbound AXIN acts as inhibitor of WNT signaling , in place of the complete destruction complex , the ( reversible ) binding states of AXIN to DVL and membrane-bound LRP6 relate to the two previously described mechanisms for activating β-catenin signaling: During the first two hours , β-catenin activation solely results from DVL/AXIN binding , i . e . the redox-dependent DVL/β-catenin pathway ., Only after that , AXIN starts getting recruited to the membrane and bound by the activated LRP6 receptor complex ., This process is driven by the auto-/paracrine WNT signaling , which , in the long run , replaces the transient redox-dependent DVL/β-catenin pathway , such that AXIN is eventually only bound to LRP6 ., Note , that due to negative feedback , the elevated concentration of nuclear β-catenin enhances the synthesis of AXIN ., As a result , in the long run , the binding of AXIN to LRP6 yields an unrestrained linear increase of LRP6/AXIN in control cells for late time points ., This indicates that additional mechanisms , like endocytosis and recycling , are required to maintain the continuous auto-/paracrine WNT-signaling for a longer period of time ( cf . Conclusion and Outlook ) ., In summary , our simulation results suggest a two-fold activation mechanism that drives the early differentiation process in human progenitor cells ., Accordingly , the cellular response upon differentiation induction through growth-factor removal is characterized by an immediate , transient response through redox-dependent DVL signaling , followed by a constant , auto-or paracrine WNT signaling in a raft-dependent manner ., In a combined in-vitro and in-silico approach we find strong evidence , that cell fate commitment in human neural progenitor cells is driven by two distinct β-catenin signaling mechanisms ., According to our simulation results , only a concisely regulated interplay between redox-dependent and self-induced auto-/paracrine WNT signaling can explain the nuclear β-catenin dynamics observed experimentally during the initial phase of differentiation: In response to growth factor removal , a transient increase of the intracellular ROS level activates DVL in a redox-dependent manner ., While DVL is primarily bound by NRX in the inactive state , ROS release the redox-sensitive association between NRX and Dishevelled ( DVL ) ., This leads to a spontanous increase of unbound DVL molecules , which immediately get activated by forming self-aggregates ., Activated DVL subsequently stimulates downstream signaling components causing an immediate transient β-catenin signal 25 , 55 ., After a certain delay , a yet unknown mechanism triggers a continuous production of WNT molecules , which results in a stable activation of WNT/β-catenin pathway by auto-/paracrine signaling ., The resulting continuous WNT signal is raft-dependent , i . e . the disruption of rafts completely inhibits the signal transduction ., Recent studies show that both WNT- and ROS-induced β-catenin signaling pathways , are essential positive regulators for the neuronal differentiation as the inhibition of either one significantly reduces the neuronal yield 5 , 21 ., In addition , we also provide a comprehensive model of WNT/β-catenin signaling that for the first time combines intracellular and membrane-related processes including lipid rafts dynamics ., Its predictive ability has been demonstrated under a wide range of varying conditions for in-vitro as well as in-silico reference data sets ., However , we are well aware , that our model is a simplified representation of WNT/β-catenin signaling ., As for instance , it does not include any endocytotic processes , like recycling or the sequestration of the destruction complex inside multivesicular endosomes as currently discussed 63 , 64 ., Though our model does neither contradict nor exclude these hypotheses ., Instead we concentrate on the fact , that phosphorylation of LRP6 is a raft-dependent process being crucial for canonical WNT/β-catenin signaling as demonstrated by 15 and our investigations ., LRP6 phoshporylation is a prerequisite for WNT-mediated endocytosis 14 , 63 ., The reversible binding of AXIN to activated LRP6 , as described in our model is sufficient to accurately predict and reproduce in-silico and in-vitro measurements under varying conditions ., However , it is one among many possible mechanisms preceeding LRP6 phosphorylation ., We’d like to emphasize that the model is based on ML-Rules , a multilevel , rule-based modelling language , that facilitates the extension and modification of the model ., With regard to the previously mentioned endocytotic processes , endosomes and multivesicular bodies ( MVB ) may thus be effortlessly included in the model in terms of dynamic , cytosolic compartments ., The presented model may thus serve as starting point to further investigate and evaluate current hypotheses referring to WNT/β-catenin signaling , like the role of raft-dependent and independent endocytosis 14 , 15 , 63 , 65 , 66 , the multiple functions of DVL in canonical and non-canonical WNT pathways ( crosstalk ) 57 , or the targeting of WNT molecules through lipid modifications 46 , 67 . | Introduction, Results/Discussion | Canonical WNT/β-catenin signaling is a central pathway in embryonic development , but it is also connected to a number of cancers and developmental disorders ., Here we apply a combined in-vitro and in-silico approach to investigate the spatio-temporal regulation of WNT/β-catenin signaling during the early neural differentiation process of human neural progenitors cells ( hNPCs ) , which form a new prospect for replacement therapies in the context of neurodegenerative diseases ., Experimental measurements indicate a second signal mechanism , in addition to canonical WNT signaling , being involved in the regulation of nuclear β-catenin levels during the cell fate commitment phase of neural differentiation ., We find that the biphasic activation of β-catenin signaling observed experimentally can only be explained through a model that combines Reactive Oxygen Species ( ROS ) and raft dependent WNT/β-catenin signaling ., Accordingly after initiation of differentiation endogenous ROS activates DVL in a redox-dependent manner leading to a transient activation of down-stream β-catenin signaling , followed by continuous auto/paracrine WNT signaling , which crucially depends on lipid rafts ., Our simulation studies further illustrate the elaborate spatio-temporal regulation of DVL , which , depending on its concentration and localization , may either act as direct inducer of the transient ROS/β-catenin signal or as amplifier during continuous auto-/parcrine WNT/β-catenin signaling ., In addition we provide the first stochastic computational model of WNT/β-catenin signaling that combines membrane-related and intracellular processes , including lipid rafts/receptor dynamics as well as WNT- and ROS-dependent β-catenin activation ., The model’s predictive ability is demonstrated under a wide range of varying conditions for in-vitro and in-silico reference data sets ., Our in-silico approach is realized in a multi-level rule-based language , that facilitates the extension and modification of the model ., Thus , our results provide both new insights and means to further our understanding of canonical WNT/β-catenin signaling and the role of ROS as intracellular signaling mediator . | Human neural progenitor cells offer the promising perspective of using in-vitro grown neural cell populations for replacement therapies in the context of neurodegenerative diseases , such as Parkinson’s or Huntington’s disease ., However , to control hNPC differentiation within the scope of stem cell engineering , a thorough understanding of cell fate determination and its endogenous regulation is required ., Here we investigate the spatio-temporal regulation of WNT/β-catenin signaling in the process of cell fate commitment in hNPCs , which has been reported to play a crucial role for the differentiation process of hNPCs ., Based on a combined in-vitro and in-silico approach we demonstrate an elaborate interplay between endogenous ROS and lipid raft dependent WNT/beta-catenin signaling controlling the nuclear beta-catenin levels throughout the initial phase of neural differentiation ., The stochastic multi-level computational model we derive from our experimental measurements adds to the family of existing WNT models , addressing major biochemical and spatial aspects of WNT/beta-catenin signaling that have not been considered in existing models so far ., Cross validation studies manifest its predictive capability for other cells and cell lines rendering the model a suitable basis for further studies also in the context of embryonic development , developmental disorders and cancers . | null | null |
journal.pcbi.1002447 | 2,012 | The Rate of Beneficial Mutations Surfing on the Wave of a Range Expansion | Our model is a variant of Kimuras stepping-stone model 18 for a population in a linear habitat , and has been used in Ref ., 4 to quantify the surfing of neutral mutations ., In this model , colonization sites ( which are called “demes” ) are regularly distributed along the axis ., Due to limited resources , each deme can only carry up to individuals ., Individuals have a certain probability to “hop” from one deme to a neighboring one ., Within one deme , logistic , stochastic growth is assumed ., Namely , if is the number of wild type individuals in a given deme , and the corresponding number of mutants , we define the corresponding ratios by and ., Then the average growth rates of wild types and mutants per unit time are given by and , respectively ., This description assumes that the individuals are haploid , but the model describes also diploids , if the fitness of the heterozygote is equal to the mean of the fitness of the homozygotes , and if is taken to mean the double of the carrying capacity of the deme ., In order to implement this model , we use a discrete algorithm , which is similar to that used by Hallatschek and Nelson 4 ., We consider a box made up by neighboring demes , and kept centered on the advancing population wave as explained below ., Each deme is filled with particles , which can be of three types: wildtype , mutant and vacancies ., ( The presence of vacancies means that the deme is not yet saturated and that the population can still grow within it . ), Then the state of the box is updated at each time step according to the following process ., Migration step: Two neighboring demes are chosen at random ., Within each of these demes , a particle is chosen at random , then those two particles are exchanged ., ( If the two particles chosen are of the same type , this leads to no change . ), Duplication step: One deme is chosen at random ., Within this deme , two particles are chosen at random , then the second particle is replaced by a duplicate of the first one , with probability ., ( Again , if the two particles chosen are of the same type , this leads to no change . ), The probability is equal to one for all processes , except for the replacement of a wild type or a mutant by a vacancy , which happen with probability and respectively ., It is possible to show that this choice indeed results in average growth rates of the form for wildtype and for mutants ., Notice that the probability that a mutant replaces a wildtype individual is equal to that of the opposite event ., Therefore , in a full deme , mutants have no competitive advantage over wildtype individuals ., However , the relative proportion of mutant and wildtype individuals will be subject to stochastic fluctuations ., We define our unit of time so that the diffusion constant of the particles is equal to one ., The results just described as well as the direct inspection of particular realizations , guided us in drafting a rough scenario for the fate of the mutants: We can now begin to provide explanations for the quantitative results of our simulations ., According to the basic scenario outlined above , mutants arising far in the tip of the wave fix depending on whether or not they avoid a stochastic death in the first stage of their growth ., Notice that the presence of a wildtype wave plays no role here , since it has not yet reached this position ., In a large well-mixed population , this survival probability is simply given by for a branching process with growth rate and death rate , which is a classical Haldane formula for the establishment probability of a beneficial mutation 19 ., This standard result remains unchanged in the present spatial model with local logistic growth , as is shown in the subsection on nondimensionalized equations by a simple argument ., Indeed , our simulations show that the probability of survival ( and fixation ) probability saturates at the value for sufficiently beneficial mutations in the tip of the population wave ., In the Results section , we defined as the typical distance , measured from the front , where the surfing probability changes from 0 to its maximal value ., In other words , mutants have very small chance to reach fixation if they are introduced at , whereas they will almost surely fix if they start at , provided that they survive the stochastic fluctuations in the first stage of their growth ., In the basic scenario described above , we suggested that this meant in fact that mutants starting further than have enough time to grow , so that they are then numerous enough to stop the advancing wildtype wave ., This argument can be turned into a quantitative estimate of the magnitude of ., According to the classical Fisher wave speed and our numerical measurements , our model ( cf ., equation ( 9 ) ) implies that the wildtype wave propagates at a velocity ., Therefore , the wildtype wave will reach the growing mutant population at a time ( where is the distance from the front to the starting position ) ., Let us now estimate how much the mutant population will have grown before this arrival time of the wild type wave ., To this end , we assume that the mutant clone grows unaffected by the wildtype population up until time ., Then , the total mutant population grows exponentially on average , according to ., However , we know from the previous subsection that , after some time , the mutant population is non-zero in only a fraction of the realizations ., Therefore , , where is the average over realizations in which the mutant population has not died out ., Thus we have ., Now we make the simple-minded assumption that the probability to fix is large if the total mutant population has grown above a characteristic number on the order of the typical number of mutants in an all-mutant wave before the wildtype wave reaches it ( i . e . , at time ) and is small otherwise ., Hence , we expect ( 2 ) where is a weakly varying function of its two arguments ., We will show in the Methods section , that indeed the only relevant parameters that govern the surfing probabilities are and , which appear as arguments in ( 2 ) ., Our estimate of the “edge” of surfing in ( 2 ) should be considered as an upper bound because a clone may not need to grow to a size as large as the number of individuals in a mutant wave front , as we have assumed in our argument ., Nevertheless , the estimate in ( 2 ) sets a useful bound on , which works very well for small populations and large fitness effects , as documented by our data in Figure 2 ., With the onset of surfing in the tip of the wave and the maximum surfing probability , we have discussed two characteristic features of the sigmoidal function ., A more detailed analysis is required , however , to describe the transition region where most of the surfing beneficial mutations are generated , which is a pre-requisite for dissecting the substitution rate below ., Therefore , we sought for a differential equation that may determine the functional form ., An equation of this kind was already found in ref ., 4 , for the case of a neutral mutation , on the basis of a backward Fokker-Planck formalism ., However , the approach that these authors use is specific to neutral mutations and cannot be extended to the non-neutral case ., For sufficiently beneficial mutations , it is however possible to derive an approximate differential equation for by employing the theory of branching random walks ., To this end , we approximate the proliferation of newly introduced mutant by a linear birth-death process: A mutant at position has a constant birth rate of per generation ., The death rate on the other hand depends on location ., Far in the tip of the wave , the death rate of the mutants approaches a constant of , and it approaches in the bulk of the wave as there is no net growth in the saturated region of the population ., By construction of our model , the net -dependent growth rate is given by , where is the number of wildtypes in a deme located at at time , and is the analogous quantity for the mutants ., Thus the net growth rate is in general fluctuating due to the fluctuating occupancy of deme ., We now make two important assumptions ., First , we assume that the survival of the mutants is decided early on when the mutant population is so small that we can well approximate its growth rate by the function , i . e . , by neglecting the non-linear effect of the mutant population on its own survival ., This approximation is justified when the growth rate advantage of the mutants is sufficiently large , and breaks down in the neutral or nearly neutral case ., Second , we average the growth rates over all realization and assume a growth rate , where is the average density profile of an all wildtype wave ., This simplification holds provided that that the carrying capacity is so large that fluctuations in the wave profile are weak ., Under these assumptions , we can use a standard result for branching random walks , namely that the survival probability , which in our case equals the surfing probability , satisfies ( 3 ) In the Methods section , we provide a heuristic rational of this differential equation , but for a strict derivation the reader is referred to standard text books , such as ref ., 21 ., Equation ( 3 ) has a form very similar to the differential equation for a deterministic Fisher-Kolmogorov wave running in the direction ., This explains the overall sigmoidal “wave profile” of the function ., Notice however that the term approaches for where the wildtype occupancy saturates , ., Thus , ( 3 ) should be regarded as a classical Fisher-Kolomgorov equation with a cut-off 22 , an observation which will be important in the following section ., To quantitatively compare the branching process theory with our individual based simulations , we integrated equation ( 3 ) numerically ., As shown in Figures 3 and 4 , the agreement is very good , and remains so when the parameters and are varied as long as ., As a proxy for the speed of adaptation at shifting range margins , we finally ask how frequently beneficial mutations fix in the pioneer population for a given mutation rate ., Clearly , the surfing probability is one important factor as it governs the chances of success for a mutation inserted at location ., We have seen that , generically , steeply increases towards the tip of the wave due to the location advantage appreciated there ., However , only few individuals reside in the tip region and can thus provide mutational input for adaptation ., This effect is described , of course , by the wave profile ., The product describes the tradeoff between the higher success probability in the tip and the higher mutational input in the bulk of the wave ., More precisely , the integral ( 4 ) controls the substitution rate for beneficial mutations of effect and mutation rate ., As argued earlier , for sufficiently beneficial mutations , the survival of a beneficial mutation is well-described by our mean-field description that only depends on the mean ., We may thus approximate by setting , and use our above results for the average survival probability and population density to estimate the integral on the right hand side ., The value of is plotted in Figure 5 as a function of the selective advantage of the mutants ., These results show that , for carrying capacities ranging from to , the substitution rate depends only weakly on selection coefficients ., Even for selection coefficients of 10% , the substitution rate in the most dense population ( ) is merely increased by a factor 4 compared to the neutral base line ., Also note , as shown in Figure 6 , that the substitution rate does increase more slowly than linear with population size ( as parameterized by ) quite in contrast to well-mixed population models ( in the absence of clonal interference 23 ) ., Our simulated data for are hard to model from first principles , as this would require a solution to the long-standing problem of noisy Fisher waves for rather small values of 20 ., However , for large carrying capacities such that , where genetic drift is weak , an analytical approach is feasible ., The analysis , described in the Methods section , not only allows us to answer the question as to how the substitution rate behaves in the deterministic limit , or relatively close to it ., It also provides us with a qualitative picture of how genetic drift , mutations and selection compete during a population expansion ., These asymptotic results are meant to guide the intuition as to how weakly selection affects the substitution process ., When a beneficial mutation arises in the front of an expanding population , it has a high risk of being immediately lost from the front population either by extinction or because the mutant clone cannot keep up with the shifting wave front ., Rarely , however , mutants become entirely fixed in the front population , a phenomenon referred to as gene surfing ., In this paper , we have studied the results of a one-dimensional individual-based simulation to measure and explain, i ) the probability of surfing of a newly introduced beneficial mutations on a population range expansion and, ii ) the rate of these surfing events if beneficial mutations occur at a certain rate and have a certain effect ., In agreement with earlier studies 4 , 6 , , we found that the probability of surfing crucially depends on the location of the first mutant with respect to the advancing wave ., We have quantified this location advantage in two ways ., First , we estimated heuristically the spatial head start required for a clone of beneficial mutations to grow large in the wave tip before the bulk of the wave arrives ., This head start was found to be inversely proportional to the growth rate of the mutants and only grows logarithmically with the carrying capacity ., If mutations arises sufficiently far ahead of the front of a population-expansion wave , they can fix even if fitness effects are small , which is consistent with earlier observations 6 , 9 , 10 ., A more systematic and accurate analysis based on the theory of branching processes could be given to describe how fast the surfing probabilities rise as one moves into the tip of the wave until it eventually saturates ., Further analysis , reported in the Methods section , shows that in the deterministic limit of infinite carrying capacities , the characteristic distance at which surfing becomes significant scales as for small selective advantage ( cf ., equation ( 23 ) ) ., For any reasonable carrying capacity , however , surfing probabilities are found to be significantly higher than expected from a deterministic analysis , which shows that genetic drift is essential for the surfing of weakly beneficial mutations ., Our analytical description of the location-dependent survival probability enabled us to get at our second key question: At what rate do surfing events occur for a given mutation rate and selective advantage ?, This rate of surfing events may be viewed a proxy for how quickly a population may evolve toward a range expansion phenotype 17 ., The surfing rate is determined by two factors ., One is , of course , the surfing probability , which increases towards the tip of the wave , the other is the mutational process by which new potential surfers are introduced ., Clearly the mutational supply is highest in the bulk of the wave because of its saturated population density , but there the surfing probability is lowest ., It turned out that , due the trade-off between both effects , most surfers are generated at an intermediate position within the front of the wave ., We were able to determine analytically the substitution rate for large populations and small mutational fitness effects ., This analysis shows that , in the deterministic limit , surfing rates for small selection coefficients are strongly suppressed ., Mathematically , this is manifest in an essential singularity of the substitution rates at vanishing selection coefficients ., For large but finite carrying capacities , however , substitution rates are strongly increased due to even tiny amounts of genetic drift ., Our theory predicts a generally quite strong positive correlation between surfing rates and genetic drift ( as quantified by inverse carrying capacities ) for small selection coefficients ., Interestingly , our simulations show that this correlation is qualitatively inverted for large selection coefficients: Very large effect mutations do not require genetic drift to prevail , so that their rate is mainly controlled by the mutational supply which increases with increasing carrying capacities ., However , our results suggests for beneficial mutations of intermediate and small effects that long-term survival during a range expansions is mostly a matter of luck to arise far in the wave tip than of fitness ., In summary , we have for the first time analyzed not only the fate of newly introduced mutations , but also the rate of surfing events for a given mutation rate ., Our results suggest that genetic drift is not required to promote mutation surfing of strongly beneficial mutations for which selection is strong enough ., Importantly , however , our results suggest that some amount of genetic drift strongly increases substitution rates at advancing fronts for weakly beneficial mutations and thus can be important for promoting adaptation towards an invasion phenotype ., Finally , we discuss the assumptions at the base of our study , and its possible generalizations ., First , we only considered mutations that are beneficial to the pioneer population but neutral for the bulk population ., Several experimental studies suggest that such mutations towards a range-expansion phenotype are actually disadvantageous in the bulk of the population 14–16 ., While such mutations gradually disappear from the bulk population , we expect that their surfing propensity will be almost identical to mutations that are neutral or beneficial in the bulk ., This is because the bulk phenotype matters so far from the wave tip that it cannot influence the genetic composition of the wave tip ., The analysis would change qualitatively if the selective advantage in the bulk is so large that the ensuing genetic wave of the beneficial mutation within the saturated bulk population would be faster than the range expansion ., However , this situation only occurs for extreme selective differences on the order of one ., We have also assumed that population expansions proceed according to R . A . Fishers standard model , in which the Malthusian growth rate of individuals in the tip of the wave is constant ., However , many species are characterized by a reduced Malthusian growth rate when densities become too small ., This effect arises when individuals need to cooperate with others in order to proliferate , for instance in the case of sexual reproduction ., Such Allee effects 24 have been found to considerably lessen the role of genetic drift in the gene surfing phenomenon: The effective population size associated with the expanding population front was strongly positively correlated with the strength of the Allee effect 4 ., We expect that such Allee effects will also alter surfing probability and rates of beneficial mutations , because they lessen the extreme location advantage of mutations arising in the far wave tip ., As a consequence , the surfing beneficial mutations arise closer to the bulk of the population for stronger Allee effects ., Also the total rate of surfing events would be strongly increased ., We thus expect that larger Allee effects will significantly enhance adaptation towards a range expansion phenotype ., Another interesting extension of our study concerns expansion waves in planar habitats ., In this case , the location advantage for deleterious mutants is likely to be less relevant , since the wildtype population is able to overcome the mutant and constrain it to a bounded region ., As in the one-dimensional case , successful long-term surfing of deleterious mutations will require that the mutant clone takes over the entire colonization front ., As a consequence , the surfing probability will sensitively depend on the habitats extension transverse to the expansion direction ., Also , the analysis of the surfing of beneficial mutations will be more complex: Surfing beneficial mutations give rise to sectors 8 with sector angles that characterize their selective advantage against the surrounding wild type population ., Furthermore , at any given time , some parts of the colonization front will be more advanced than others , due to the inevitable random front undulations ., If a mutation arises in one of those more advanced region of the habitat , it will have higher long-term surfing probabilities than in the less advanced regions ., Nevertheless , simulations of the kind carried out in this study should quite generally allow to investigate the establishment probabilities in any model of expanding populations ., In the present subsection we show that the different parameters , , and which define the model enter in fact only in the combinations and ., In particular , this explains the behavior of shown in Figure 2 ., In order to do so , we recast the dynamics of the model in terms of stochastic differential equations ., Let us denote by ( ) the position of the deme ., Then the state of the system is identified by the -dimensional vector ., Thus the algorithm described in the previous section can be represented by a master equation of the form ( 5 ) where the index runs over all the allowed types of events that lead to a change in ( birth , death , migration to a neighboring deme , etc . ) , is the probability of such an event per unit time and is the resulting variation of the vector ., The expressions of and for each allowed event are detailed in Table 1 ., Expanding equation ( 5 ) to first order in ( see , e . g . , 26 , chap . ∼X ) leads to a Fokker-Planck equation , and a corresponding set of Langevin equations can then be found ., Under the assumptions that and , we may approximate the and by the continuous functions and ., If we further assume that , and that stochastic deviations from the average diffusion term are negligible , these equations read: ( 6 ) ( 7 ) In this expression , the Gaussian noises , and are uncorrelated , and one has , for instance , ., This set of equations corresponds to a stochastic reaction-diffusion system , where the reaction term is logistic , and where , by construction , the diffusion constant is equal to 1 ., Notice that the last term corresponds to the stochastic replacement of a mutant by a wildtype individual ( or conversely ) and is responsible for stochastic fluctuations within a full deme ., The equations can be made nondimensional by setting ( 8 ) We obtain therefore ( 9 ) ( 10 ) The nondimensionalized equations reveal , as anticipated , that the problem only depends on two relevant parameters: and ., The survival probability of a linear branching process with birth rate and death rate can be easily determined by the following discrete reasoning: let us denote the total number of individuals by , and consider the probability that a population of individuals will survive ., Diffusion events do not change ; it is only affected by duplication events ( births or deaths ) ., However , death events are always times less likely than birth events ( see the definition of the model ) ., Thus , a given duplication event is a birth with probability and a death with probability ., By conservation of the probability after such an event we have ( 11 ) with the boundary conditions and ., We obtain therefore ( 12 ) Thus the probability that the population stemming from one single mutant will survive is given by ( 13 ) In the bulk , the only possible events are the replacement of a mutant by a wildtype individual ( or the opposite ) , which take place with the same probability ., Thus the size of an isolated mutant population in the bulk undergoes a critical branching process in the presence of an infinite reservoir of wildtype individuals , and its survival probability vanishes ., Here , we provide a heuristic rational for the differential equation ( 3 ) for the surfing probability ., Let us consider the introduction of a mutant at time and position ., We denote the probability to find a mutant at a position and at a later time by ., Now , let us place ourselves in the conditions in which In this case , if we find a mutant at and , its situation is essentially the same as if it had just been introduced in a wave consisting only of wildtype individuals , since and since , if there are other mutants in the wave at , they will probably not perturb its dynamics ., Indeed , for small , other mutants will disappear , in most realizations , before getting a chance to interact effectively with the mutant we consider ., Therefore , for this mutant at and , the probability to fix is by definition ., We may therefore decompose the probability as follows: ( 14 ) However , this formula is an overestimate of ., Rare realizations in which two mutants are present at , and in which the issues of both survive , should be counted as one single fixation event , but are in fact double-counted by the formula ( 14 ) ., Therefore , we expect a negative correction of order when becomes larger ., If , however , we neglect for the moment this correction , differentiating equation ( 14 ) with respect to leads to ( 15 ) Notice that , in fact , ., Since the mutant population is not very large at , we can neglect the term in equation ( 7 ) , and replace by ., Therefore , in the frame moving with the velocity of the wave , equation ( 7 ) becomes , for : ( 16 ) Upon substituting this expression of in equation ( 15 ) , integrating by parts , and noticing that the equation is valid for all , we obtain the necessary condition ( 17 ) Because of the assumptions that were used in its derivation , this equation is only valid when is small , i . e . , close to the bulk of the wave ., However , far ahead of the front , equation ( 17 ) does not predict the observed saturation of at ., We attribute this to the fact that we neglected corrections of order ., Therefore , we may add a phenomenological non-linear term to equation ( 17 ) : ( 18 ) This term leaves equation ( 17 ) unchanged when is small , but leads to the correct saturation at far from the front ., Our analysis of the substitution rate starts from the observation that the integrand in the expression for the substitution rate in equation ( 4 ) has mainly support in the region where decays exponentially , and increases exponentially , see Figure 1 ., This reflects the tradeoff between high population density ( required for the production of mutations ) and high surfing probability ( required for the fixation of mutations ) that determines the substitution process ., In the regions that significantly contribute to , we may thus approximate the wild type wave profile by ( 19 ) for and otherwise ., Here , is the actual speed observed for the wild type wave ., Secondly , we approximate by ( 20 ) for , and otherwise ., Here , is the deterministic speed of a mutant wave ., Using these exponential approximations , we can estimate as ( 21 ) Equation ( 21 ) is hard to evaluate for general and selection strength ., However , one can derive an asymptotically correct expression for in the limit of large for fixed , where the exponential approximation is the leading order description of the wave profile 22 ., In this limit , the equation for the survival probability describes a Fisher wave running in the direction with a cutoff far in the tip of the wave , as discussed after Eq ., ( 3 ) ., The cutoff ( due to the net growth rate being proportional to in ( 3 ) ) has the effect of lowering the wave speed from the deterministic value to the wildtype value ., With the cut-off at position one obtains an asymptotic wave speed of 22 ., For this lowered wave speed to equal the wildtype speed , we find ( 22 ) ( 23 ) where the last equation holds for ., Since we have in the limit , we can now express ( 21 ) in terms of our model parameters , obtaining ( 24 ) which holds for small ., Notice that is characterized by an essential singularity for , which causes very small substitution rates for small , indicating that selection is very inefficient at advancing fronts ., Our analysis neglected so far the effects of a finite carrying capacity ., We can account for finite to leading order by taking advantage of known results for noisy traveling waves , i . e . , the fact that to leading order is given by 22 ( 25 ) Inserting this expression in ( 22 ) yields a substitution rate of ( 26 ) In Figure 7 , we plot the theoretical predictions for vs . , while simulation data are shown in Figure 6 ., Notice that the expression with finite stays far below the deterministic limit for any reasonable value of , which results in a non-trivial power law dependence on ., From the expression in ( 26 ) , it is clear that the effect of a finite carrying capacity is important unless , which requires extremely large populations for reasonable selection coefficients ., In the opposite quasi-neutral case , the expression for lead reduces to the position of the cutoff in a noisy Fisher wave , 22 . | Introduction, Results, Discussion, Methods | Many theoretical and experimental studies suggest that range expansions can have severe consequences for the gene pool of the expanding population ., Due to strongly enhanced genetic drift at the advancing frontier , neutral and weakly deleterious mutations can reach large frequencies in the newly colonized regions , as if they were surfing the front of the range expansion ., These findings raise the question of how frequently beneficial mutations successfully surf at shifting range margins , thereby promoting adaptation towards a range-expansion phenotype ., Here , we use individual-based simulations to study the surfing statistics of recurrent beneficial mutations on wave-like range expansions in linear habitats ., We show that the rate of surfing depends on two strongly antagonistic factors , the probability of surfing given the spatial location of a novel mutation and the rate of occurrence of mutations at that location ., The surfing probability strongly increases towards the tip of the wave ., Novel mutations are unlikely to surf unless they enjoy a spatial head start compared to the bulk of the population ., The needed head start is shown to be proportional to the inverse fitness of the mutant type , and only weakly dependent on the carrying capacity ., The precise location dependence of surfing probabilities is derived from the non-extinction probability of a branching process within a moving field of growth rates ., The second factor is the mutation occurrence which strongly decreases towards the tip of the wave ., Thus , most successful mutations arise at an intermediate position in the front of the wave ., We present an analytic theory for the tradeoff between these factors that allows to predict how frequently substitutions by beneficial mutations occur at invasion fronts ., We find that small amounts of genetic drift increase the fixation rate of beneficial mutations at the advancing front , and thus could be important for adaptation during species invasions . | When a life form expands its range , the individuals close to the expanding front are more likely to dominate the gene pool of the newly colonized territory ., This leads to the sweeping of pioneer genes across the newly colonized , a process which has been named gene surfing ., We investigate how this effect interferes with natural selection by evaluating the probability that an advantageous mutant , appearing close to the edge of an advancing population wave , is eventually able to dominate the population range expansion ., By numerical simulations and heuristic analysis , we find that the surfing of even strongly beneficial mutations requires that they are introduced with a certain spatial head start compared to the bulk of the population ., However , as one moves ahead of the wave , one finds fewer and fewer individuals which can possibly mutate ., As a consequence , successful mutations are most likely to arise at an intermediate position in front of the wave ., For small selective advantage , the success probability is enhanced by an even smaller amount of genetic drift ., This effect could be important in aiding adaptation to local conditions in a range-expansion process . | genetics, population genetics, biology, evolutionary biology, population biology, evolutionary processes, genetics and genomics | null |
journal.ppat.1002302 | 2,011 | Trafficking of Hepatitis C Virus Core Protein during Virus Particle Assembly | Hepatitis C virus ( HCV ) is a major cause of acute and chronic hepatitis , cirrhosis , and hepatocellular carcinoma ., HCV is an enveloped , positive-strand RNA virus classified with the Family Flaviviridae 1 ., The viral genome encodes an open reading frame of ≈3011 codons that is translated as a single polyprotein , which is cleaved by viral and host proteases into at least 10 distinct products ( Figure 1A ) ., The N-terminal region encodes three structural components: core protein , which forms the viral nucleocapsid , and two envelope glycoproteins ( E1 and E2 ) , which mediate viral attachment and entry ., The remainder of the genome encodes the nonstructural ( NS ) proteins: p7 , NS2 , NS3 , NS4A , NS4B , NS5A and NS5B ., The NS proteins mediate intracellular aspects of the virus life cycle including RNA replication , subversion of innate antiviral defense , and virus particle assembly ., The precise roles of NS proteins in virus particle assembly are not clear but p7 , NS2 , NS3 , NS4A , NS4B , and NS5A all contribute to this process 2 , 3 , 4 ., HCV core is a highly basic RNA-binding protein that contains three distinct functional domains 5 ., Domain 1 ( amino acid ( aa ) 1–117 ) is hydrophilic and contains determinants for RNA binding and core oligomerization 6 ., Domain 2 ( aa 118–177 ) forms a pair of amphipathic helices that mediate the peripheral association of core with cellular membranes 7 , 8 , 9 ., Domain 3 ( aa 178–191 ) , which serves as a signal peptide for the translocation of E1 protein into the endoplasmic reticulum ( ER ) lumen , is absent from mature core protein 5 ., Core is initially cleaved from the polyprotein by host signal peptidase ( SP ) ; subsequent removal of domain 3 by signal peptide peptidase ( SPP ) then yields mature core protein that forms a homodimer 6 , 9 , 10 ., Following cleavage , mature core protein is targeted to lipid droplets ( LDs ) 11 , 12 , 13 ., LDs are intracellular storage organelles containing a hydrophobic core of neutral lipids and cholesterol esters surrounded by a phospholipid monolayer embedded with LD-specific proteins 14 ., LD biogenesis is not fully understood , but LDs are likely derived from the outer leaflet of the ER and may remain contiguous with this membrane system 15 ., LD-associated proteins are presumably loaded onto LDs at sites of ER contact 16 , although vesicular transport mechanisms have not been formally excluded ., The best-characterized LD-associated proteins are perilipin , adipocyte differentiation-related protein ( ADRP ) , and tail-interacting protein ( TIP ) 47 , collectively known as the PAT proteins 17 ., PAT proteins are thought to regulate the dynamics of lipid acquisition , storage , and release 15 ., In addition , the membrane trafficking GTPase Rab18 may associate with a subset of LDs undergoing lipolysis 18 , 19 , 20 , 21 ., The role of core trafficking to LDs is not well understood ., Prior work has shown that core protein recruits NS proteins and RNA replication complexes to sites adjacent to LDs 22 ., Furthermore , core recruits NS5A to the surface of LDs , where they co-localize 23 ., Mutations that alter the LD localization of core or that block cores ability to recruit viral NS proteins to LDs inhibit virus production 22 , 24 , 25 , 26 , suggesting that LDs are intimately involved in virus particle assembly ., The site of virus budding has not been definitively determined , but the ER retention of E1-E2 27 , 28 , the complex glycan modifications on secreted virus particles 29 , the differential effects of Brefeldin A ( BFA ) on virus assembly vs . virus secretion 30 , and analogies to closely related flaviviruses ( reviewed in 31 ) , all suggest that virus particles bud into the ER and transit through the secretory pathway ., However , it is not yet clear how LD-associated core contributes to this process ., We hypothesize that core protein must be trafficked from LDs into nascent virus particles at the LD-ER interface ., To understand the dynamics of core protein trafficking during virus assembly , we developed methods to fluorescently label and image functional core protein in live , virus producing cells ., We observed core trafficking to static ADRP-positive LDs , forming a cap on the surface ., At these sites , core co-localized with the viral E2 glycoprotein and adjacent to NS3 protein , consistent with these being sites of virus assembly ., We also observed highly motile ADRP-independent core puncta that represent post-LD form of core ., By using pharmacologic inhibitors of virus egress and a panel of mutants blocked in virus assembly , we showed that core is recruited from sites of assembly into these puncta , and that this process requires interaction between viral NS proteins ., To better understand the trafficking of HCV core protein during virus particle assembly , we developed methods to fluorescently label and image functional core in living cells ., Traditional live imaging systems often rely on the insertion of fluorescent proteins into a target protein , but the relatively large size of such tags would likely interfere with the function of core protein ., We therefore genetically inserted the small tetracysteine ( TC ) peptide tag ( FLNCCPGCCMEP ) near the N-terminus of core ( Figure 1B ) within the context of the HCV Jc1 infectious clone ., Importantly , insertion of this tag had only minimal effects on infectious virus production compared to untagged Jc1 ( Figure 1C ) ., For both untagged Jc1 and Jc1/core ( TC ) , peak viral infectivity was observed between 48 and 72 h post-electroporation , although the infectivity titers of Jc1/core ( TC ) were slightly reduced ( 2- to 4-fold ) at each time point ., This correlated with a 2 . 7-fold decrease in specific infectivity ( 0 . 16 Jc1 infectious units per RNA-containing particle vs . 0 . 06 Jc1/core ( TC ) infectious units per RNA-containing particle ) , suggesting that the small decrease in infectivity titers may have been due to an inefficiency in virus entry rather than in virus assembly ., Both Jc1 and Jc1/core ( TC ) had similar biophysical profiles , with peak infectivities and specific infectivities observed in fractions with buoyant density of ≈1 . 10 g/ml ( Figures S1A and S1B ) ., Furthermore , the TC-tagged form of core accumulated to similar levels as untagged core within virus-producing cells ( Figure 1D ) , and the TC insertion was retained after six serial virus passages ( data not shown ) ., Together , these data indicated that the TC tag insertion was well tolerated and that the TC-tagged core protein was functional for virus assembly ., To label core ( TC ) , infected Huh-7 . 5 cells were incubated with FlAsH ( green ) or ReAsH ( red ) under optimized labeling conditions ( as described in Materials and Methods ) during the peak of virus assembly , 48 to 72 h post-infection or electroporation ( Figure 1C ) ., As shown in Figure 1E , specific signals were observed in Jc1/core ( TC ) -infected cells , often as bright puncta or crescents ( Figure 1E , arrowhead ) , but not in Jc1-infected cells ., Importantly , these FlAsH labeling conditions had no effect on the release of infectious virus particles ( Figure 1F ) , indicating that these methods allowed us to label functional core protein during virus assembly ., To further confirm that FlAsH-labeling was specific for core , Jc1/core ( TC ) -infected cells were labeled with FlAsH , fixed , and stained for core by IF ( Figure 1G ) ., FlAsH and anti-core IF signal largely co-localized ( Pearsons correlation coefficient\u200a=\u200a0 . 715 ) , confirming that FlAsH labeled core ( TC ) protein ., However co-localization was incomplete , largely due to the increased signal intensity of the IF signal in a perinuclear , reticular pattern ., This difference could be attributable to one or more of the following issues ., First , FlAsH signal intensity decreased 4-fold during fixation and processing for IF ( Figure S2 ) , mostly likely because this dye lacks an aldehyde-reactive primary amine and may be washed out during processing ., Second , indirect IF is designed to amplify weak signals , whereas FlAsH labeling binds stoichiometrically to the TC tag ., Third , FlAsH is less photostable than the Alexa dye used during IF , which could bias the relative signal intensities ., Fourth , small dyes and antibodies may differ in their accessibility and efficiency of protein labeling ., Nevertheless , these data confirmed that the TC tag can be used to label core protein in live , virus-producing cells ., To better clarify the role of core LD-trafficking during virus assembly , we created Huh-7 . 5-derived cell lines that stably expressed ADRP , a marker of storage LDs , fused to green fluorescent protein ( GFP ) or cerulean fluorescent protein ( CFP ) ., As ADRP expression levels can influence LD metabolism 32 , we first characterized these cell lines for their ADRP expression , LD content , and ability to support infectious HCV particle assembly ., Neither tagged-ADRP protein was overexpressed when compared to endogenous levels of ADRP expression ( Figure S3C ) ., There was a modest increase in the number of LDs in Huh-7 . 5/CFP-ADRP cells ( 170 . 8±67 . 89 LDs/cell ) compared to Huh-7 . 5′s ( 128 . 0±53 . 1 LDs/cell ) , although this difference was not statistically significant ( p >0 . 05; unpaired Students t-test ) ., Furthermore , there was no difference in the volume of LDs , either with or without oleic acid supplementation ( Figure S3D ) ., Importantly , both cell lines supported infectious virus production at levels comparable to non-transduced cells ( Figure S3A–B ) ., These data indicated that the GFP-ADRP and CFP-ADRP cells provided suitable environments to study core-LD trafficking in live cells ., To investigate core trafficking , we performed ReAsH labeling and live cell imaging of core ( TC ) in Huh-7 . 5/GFP-ADRP cells ., Across multiple experiments , core consistently localized to:, i ) a dim reticular pattern , most likely the ER ( Figure S3E ) 13 , 22;, ii ) caps on ADRP-positive LDs; and, iii ) bright puncta that were not associated with ADRP-positive LDs ( Figure 2A ) ., Notably , the caps of LD-associated core protein frequently faced the ER-like form of core , and core-positive LDs exhibited little directional movement ( Figure 2B , upper panel and Video S1 ) ., In contrast , ADRP-positive LDs that lacked core were more motile ., Furthermore , core puncta were small and motile ( Figure 2B , lower panel and Video S1 ) ., Given the velocity of core puncta , we examined whether they were associated with microtubules by performing live cell imaging in Huh-7 . 5 cells that stably expressed fluorescent protein-tagged ß-tubulin ., As shown in Video S2 , core puncta trafficked along microtubules , frequently in a retrograde direction ., In contrast , motile core puncta were not associated with fluorescent protein-tagged actin filaments ( data not shown ) ., Furthermore the motility of core puncta was inhibited by nocodazole , an inhibitor of microtubule trafficking ( Figure 2C and Video S3 ) ., These data indicated that motile core puncta were trafficked on microtubules ., To better clarify the interaction of core with LDs , and to determine whether core puncta represent small LDs that lack ADRP , we examined additional LD markers ., Staining with a dye specific for neutral lipids confirmed that core localized to semi-spherical caps on the surface of LDs and to small LD-independent puncta ( Figure S3F ) ., Furthermore , LD-associated core specifically trafficked to ADRP-positive LDs but not to Rab18-positive LDs ( Figures S3G and H ) , which likely represent LDs undergoing lipolysis 33 ., These data confirmed that core trafficked to ADRP-positive LDs and to distinct motile puncta that were not LD-associated ., To determine whether the different forms of core corresponded to sites of virus particle assembly , we examined the localization of core with respect to the viral E2 glycoprotein ( a structural component of virus particles ) or NS3 serine protease-RNA helicase ( a NS protein that has been implicated in virus assembly ) ., Since we currently lack tools to image functional E2 and NS3 in live cells , these viral proteins were localized by IF in fixed cells ., As seen in Figure 2D , the majority of E2 staining was in a reticular pattern , consistent with its ER retention , as well as discrete E2 puncta that co-localized with core ( Figure 2D , arrowheads ) ., These core-E2 structures were frequently found adjacent to small LDs and may represent areas where core has been recruited from LDs into nascent virus particles ., The frequency of such core-E2 puncta was low ( 2 . 86±0 . 54 per cell , n\u200a=\u200a28 cells ) , which could reflect the low burst size of HCV as well as the inherent difficulties with antibody recognition of E2 34 ., In addition , we also observed core puncta that were not labeled with E2 ., Similarly , NS3 also showed a reticular staining pattern , but frequently concentrated in regions adjacent to LD-associated caps of core protein ( Figure 2E , arrowheads ) ., For further characterization , we looked for co-localization of core with other components of the secretory compartment , including markers for ER exit sites ( Sec16 , Sec23 ) , the ER-Golgi intermediate compartment ( ERGIC-58 ) , and Golgi ( TGN38 ) , but did not reproducibly observe co-localization with these markers ( data not shown ) ., Based on these data , we hypothesize that LD-associated core caps likely represent sites of early virus particle assembly , while motile puncta may represent core-containing transport vesicles ., To further clarify the relationship between core trafficking and virus particle assembly , we examined core trafficking in a mutant virus , D88 , which is blocked at an early stage of virus particle assembly due to a large in-frame deletion in the E1-E2-p7 genes 35 ., The Jc1/core ( TC ) D88 mutant showed a substantial increase in core accumulation on the surface of LDs ( Figures 2F , S3I ) ., While a small number of core puncta were observed , they were non-motile ( Video S4 ) , indicating that they were distinct from the motile core puncta seen in virus-producing cells ., While it is not yet clear what these non-motile core puncta represent , these data showed that core accumulates on LDs and that motile core puncta were not seen when virus assembly was blocked at an early step ., Our previous results suggested that motile core puncta may represent a post LD-form of core ., To further investigate this hypothesis , we treated cells with BFA , a fungal metabolite that disrupts ER-Golgi trafficking by inhibiting the activation of ADP ribosylation factor 1 ( ARF1 ) 36 ., The timing and dose of BFA treatment were chosen to minimize effects on HCV RNA replication 37 ., Consistent with previous findings 38 , BFA treatment inhibited the secretion of virus particles as well as a model secreted protein , causing them to accumulate within BFA-treated cells ( Figure S4A–B ) ., As ARF1 can regulate lipid homeostasis 39 , 40 , 41 , we also checked whether our conditions affected LD trafficking ., Our BFA treatment conditions had only minimal effects on ADRP protein expression , LD number , and LD volume ( Figure S4C ) , indicating that these conditions could be used to study the trafficking of core protein when ARF1 is inhibited ., To observe the effects of BFA on core localization , cells were imaged before and after BFA treatment , as well as after drug washout ( Figure 3A ) ., BFA treatment increased both the number of LDs containing core and the amount of core that accumulated on each LD ( Figure 3B–D ) , suggesting a defect in core egress from LDs ., After drug washout , both of these effects were relieved ( Figure 3B–D ) ., BFA treatment resulted in a slight reduction of LD-independent core puncta at 4 h ( Figure 3E ) ; however , these puncta were not motile when BFA was present ( Video S5 and Figure 3I ) ., Upon BFA washout , more core puncta were observed ( Figures 3E , 3H ) and core puncta quickly regained their motility ( Video S6 and Figure 3I ) ., These effects were mirrored in the intracellular accumulation of virus during BFA treatment and an increase in virus secretion after BFA washout ( Figures 3F and 3G ) ., The absence of motile core puncta during BFA treatment and their re-emergence upon washout suggests that they represent a post-LD form of core ., Based on the above results , we expected to observe the trafficking of core from LDs into motile core puncta during live imaging studies ., However , these events may be relatively rare , and FlAsH and ReAsH are prone to photobleaching , which greatly limited the number of sequential frames that could be acquired during time course experiments ., To better monitor core trafficking over time , we took advantage of the dual labeling capabilities of biarsenical dyes 42 ., We reasoned that infected cells could be sequentially labeled under pulse-chase conditions , first with FlAsH to label pre-existing ( “old” ) core , followed by ReAsH to specifically detect newly synthesized ( “new” ) core ., Pilot experiments showed that simultaneous labeling of Jc1/core ( TC ) -infected cells with both FlAsH and ReAsH yielded green and red signals that completely overlapped , indicating that both dyes bind TC-tagged core with similar efficiency ( data not shown ) ., Next , infected cells were labeled with FlAsH , then labeled with ReAsH after appropriate intervals ( Figure 4A ) ., Under these conditions , newly synthesized protein was specifically detected with ReAsH , but not when protein synthesis was halted by cycloheximide treatment during the chase period ( Figure 4B , compare second and third rows ) ., Furthermore , newly synthesized core did not traffic to LDs when the maturation of core was blocked by treatment with an inhibitor of signal peptide peptidase but was restored after washout of this inhibitor ( Figure S5 ) ., These data confirmed that the dual labeling technique could be used to specifically label and image old and new core under pulse-chase conditions ., We used dual labeling to observe the trafficking of core synthesized before and after chase periods of 2 , 8 , and 24 h ( Figure 4B ) ., After 2 h , a small amount of newly synthesized core was detected in an ER-like reticular pattern or co-localized with old core in LD-associated caps ( Figure 4B , top row ) ., By 8 h , new core and old core had accumulated on LDs to comparable levels , and mixed puncta ( containing both old and new core ) were abundant ( Figure 4B , second row ) ., By 24 h , new core was the predominant species in both LDs and motile core puncta ( Figure 4B , bottom row ) ., Similar results were also obtained when the order of FlAsH- and ReAsH-labeling were switched ( data not shown ) ., In order to observe the trends of core trafficking , we quantitated LDs and motile core puncta that contained old , new , or mixed core protein over time ( Figure 4C ) ., These data yielded several interesting results ., First , newly synthesized core was targeted to LDs shortly after synthesis ., Second , the mixing of old and new core on LDs at 8 h indicated that core-containing LDs maintain communication with the site of core synthesis for extended periods of time ( see Discussion ) ., Third , the peak of mixed core in puncta ( 8, h ) was observed after the peak of mixed core on LDs ( 2, h ) , further supporting our hypothesis that motile core puncta represent a post-LD form of core ., Fourth , the small proportion of motile puncta containing old and mixed core at 24 h suggested that once core leaves LDs , it is either packaged into virus particles for secretion , or turned over ., We next examined the role of NS2 in core trafficking ., Prior genetic and biochemical studies indicated that NS2 plays an important role in virus assembly by bringing together the viral E1–E2 glycoprotein and NS3-4A enzyme complexes 4 , 43 , 44 , 45 ., We previously identified two classes of NS2 mutants with defects in virus assembly 45 , 46 ., Class 1 mutants ( NS2 K27A , W35A , or Y39A ) show reduced interaction between NS2 and NS3 , and their defects in virus assembly can be suppressed by a second-site mutation in the helicase domain of NS3 , Q221L 45 , 46 ., A class 2 mutant ( NS2 K81A ) has normal levels of NS2–NS3 interaction but reduced interaction between NS2 and E1–E2 , and its defect in virus assembly can be suppressed by a second site mutation in E1 , E78T 45 , 46 ., When the class 1 NS2 mutation K27A was introduced into Jc1/core ( TC ) , the amount of core staining per cell increased , and specifically , core accumulated on LDs ( Figures 5A and 5B ) , similar to what was previously observed with the D88 mutation ., The addition of the NS3 Q221L suppressor restored normal core trafficking ( Figures 5A and 5B ) ., Based on these results , we tested additional NS2 mutants ., All class 1 NS2 mutants showed increased intracellular accumulation of core ( Figure 5C , left panel ) , and specifically , core localized to LDs ( Figure S6A ) ., Furthermore , core accumulation was restored to WT levels by the NS3 Q221 suppressor mutation ( Figure 5C , right panel ) ., In contrast , the class 2 NS2 mutant , K81A , did not show these effects ( Figure 5C ) ., These results were further confirmed by core IF with the untagged Jc1/NS2 ( K27A ) mutant in non-transduced Huh-7 . 5 cells ( Figure S6B ) ., Taken together , these data indicated that the genetic interaction between NS2 and NS3 is important for proper egress of core from the surface of LDs ., To further clarify whether the interaction between NS2 and NS3 is important for the recruitment of core from LDs into core puncta , we performed FlAsH/ReAsH-dual labeling and quantitated LD-associated core and motile core puncta between 48 and 72 h post-electroporation ., The NS2 Y39A mutant was used for these experiments because it showed the greatest accumulation of core ( Figure 5C ) ., At all time points , the Y39A mutant showed an abundance of LDs containing old core and few LDs containing mixed or new core ( Figure 5D , left panel ) ., In contrast , the Y39A + Q221L double mutant showed LDs containing a mixture of old , new , and mixed core ( Figure 5E , left panel ) , which was qualitatively similar to WT ( Figure 4C , top panel ) ., Importantly , the number of motile core puncta , especially those containing new core , was significantly reduced in the Y39A mutant ( Figure 5D , right panel ) but was restored in the Y39A + Q221L double mutant ( Figure 5E , right panel ) ., Taken together , we conclude that the interaction between NS2 and NS3 is important for the recruitment of core from LDs into motile core puncta ., We developed methods to image functional core protein in live , virus-producing cells ., The small size of the TC tag offered several advantages over fluorescent protein tagging , namely that tagged proteins are more likely to retain native function , and do not require lengthy maturation of protein-encoded fluorophores ., Additionally , live cell imaging bypasses the need for fixation , which can alter the localization of LD-associated proteins 47 ., In the current study , TC-tagging and labeling with biarsenical dyes had minimal effects on viral replication , core protein expression , or virus particle production ., Furthermore , the optimized labeling conditions used here 48 were specific for TC-tagged core protein ., Thus , these methods can be used to confidently track functional core protein ., Despite repeated attempts , we did not observe FlAsH-labeled extracellular virus particles ., This could be explained by several considerations ., First , FlAsH label may be inefficiently incorporated into virus particles , perhaps displaced from the TC tag during RNA packaging ., Alternatively , incorporated FlAsH label could be quenched or too dim to image reliably ., The detection of extracellular virus particles may also be hampered by the low burst size of HCV 34 , the inherent pulse-type labeling with biarsenical dyes , and the propensity of these reagents to photobleach ., Nevertheless , we were able to reliably image intracellular forms of HCV core protein ., Three forms of core protein were observed in live cells: ER-associated core , LD-associated core , and motile , LD-independent puncta ., On LDs , core formed polarized caps that displaced ADRP and were frequently in close apposition to ER-localized core ., Similar LD-associated caps of core protein were previously seen in fixed cells by IF staining 24 , 49 , and likely reflect sites of core protein transfer between the ER and LDs ., Consistent with this model , newly synthesized core was directed to LDs that already contained older core ., These data suggest that core-containing LDs maintain direct communication with the ER for extended periods of time or that LDs containing old or new core can fuse to allow mixing ., Large core-containing LDs were immotile during the time periods we observed , which coincided with the peak of virus particle assembly ., Prior studies have shown that infection with HCV strain JFH-1 or overexpression of core protein causes microtubule-based trafficking of LDs to the perinuclear region 49 , 50 ., In our hands , attempts to image core at significantly earlier times were unfruitful due to low levels of core expression and dim staining ., Nevertheless , we did observe core-containing LDs clustered in the perinuclear region , suggesting that they had either formed there or moved there prior to staining ., Although LD-associated core has been implicated in virus particle assembly , it has been difficult to demonstrate a direct role for core-LD trafficking in this process ., Boulant and colleagues proposed that progressive trafficking of core onto LDs strongly correlates with a rise in infectious virus production 24 ., Furthermore , the accumulation of core on LDs inversely correlates with the efficiency of virus production , exemplified by the different localization patterns observed for viruses with high ( Jc1 ) and low ( JFH1 ) infectious titers 22 , 26 ., Similarly , we observed dramatic increases in the amount of LD-associated core with an HCV deletion mutant that is unable to assemble virus particles ., In addition to LD-associated core , we also observed motile core puncta that were not LD-associated ., We propose that motile core puncta represent a form of core relevant for virus assembly based on the following considerations:, 1 ) the formation of motile core puncta was blocked in the D88 mutant , which is unable to assemble virus particles ( Figure 2F and Video S4 ) ;, 2 ) the formation of motile puncta after BFA washout indicated that they represent a post-LD form of core ( Figure 3 ) ; and, 3 ) newly synthesized core trafficked to LDs before it trafficked to motile puncta ( Figure 4 ) ., These data suggest that puncta represent transport vesicles containing virus particles or an intermediate in virus particle assembly ., For instance , Lai and colleagues recently showed that core may traffic to a compartment containing early endosomal markers during virus particle secretion 51 ., It was interesting that core accumulated on the surface of LDs during BFA treatment ( Figure 3B–D ) , suggesting that ARF1 activity is required for the egress of core from LDs ., This would be consistent with a defect in virus assembly , as was seen for the D88 mutant ( Figure 2F ) and the class 1 NS2 mutants ( Figure 5 ) ., While BFA inhibited virus secretion , intracellular virus particles still assembled during treatment ( Figure 3F–G and S4A–B ) ., One possibility is that virus particles made during BFA treatment may utilize the available ER-associated pool of core; presumably then BFA would cause a defect in virus assembly once this pool is depleted ., In addition , BFA may slow the egress of core from LDs and reduce the rate of virus assembly ., A full accounting of infectious virus production could be used to discriminate between these possibilities ., Taken together , our data support the model for core trafficking depicted in Figure 6A ., Core protein is synthesized on the ER and trafficked to LDs , which remain in direct or indirect communication with the ER ., During virus assembly , core protein is recruited from the surface of LDs and into puncta , which likely represent transport vesicles containing virus particles or possibly a core-containing intermediate ., To further explore our model of core trafficking , we utilized our live-cell imaging tools to study the role of NS2 in virus assembly ., The class 1 NS2 mutants all showed a large accumulation of core around LDs and fewer core puncta , and core trafficking was restored by the NS3 Q221L second-site suppressor mutation in the viral RNA helicase ., In contrast the NS2 K81A mutant showed normal core-LD trafficking ., This mutant was previously found to have a distinct defect in virus particle assembly , exhibits normal levels of interaction between NS2 and NS3-4A 45 , and is not rescued by the NS3 Q221L mutation 46 ., By using sequential labeling , we showed that a block in virus assembly caused an accumulation of old core on LDs and a vast reduction in newly synthesized core on LDs ., Thus , the assembly defect results in either reduced synthesis of core , or increased turnover ., Importantly , there was a distinct lack of puncta containing newly synthesized core ., Taken together , these data strongly suggest that NS2 , and more specifically , the interaction between NS2 and NS3-4A , is important to recruit core off the surface of LDs and into core-containing puncta ., Although NS2 is important for the recruitment of core from LDs , NS2 does not directly interact with core protein 43 , 44 , 45 ., How then does NS2 contribute to core trafficking ?, NS2 coordinates virus particle assembly by bringing together the viral structural and NS proteins ( Figure 6B ) ., We propose that the interaction with NS2 may signal to NS3-4A that it is time to stop replicating viral genomes and to start packaging them ., Thus , the NS2-NS3 interaction would supply RNAs for packaging; in the absence of proper interaction between NS2 and NS3-4A , core may accumulate on LDs due to a lack of viral RNA for packaging ., In summary , we have developed methods to image functional core protein in live cells ., This system allows us to study trafficking of core protein in virus-producing cells , and revealed several novel aspects of core trafficking , including a role for interaction between NS2 and NS3-NS4A in the egress of core from the surface of LDs ., Plasmids pJc1 and pJc1/Gluc2A were recently described 46 ., Jc1/core ( TC ) was constructed in multiple steps ., First , the Jc1 5′ noncoding region ( NCR ) -core junction was amplified by using KlenTaq-LA ( DNA Polymerase Technology , St . Louis , MO ) with oligos YO-0021 ( 5′-AGA CCG TGC ACC ATG AGC TTT CTC AAT TGT TGT CCT GGC TGT TGT ATG GAA CCT AGC GGA TCC ACA AAT CCT AAA CC-3′ ) and YO-0028 ( 5′-CCA CGT GCA GCC GAA CCA-3′ ) ., The amplicon was cloned into pCR2 . 1-TOPO ( Invitrogen , Carlsbad , CA ) and sequenced ., The modified 5′ NCR-core junction was then subcloned as a 767-bp ApaLI/AgeI fragment into pJc1 46 by using common restriction sites ., pJc1/core ( TC ) D88 was constructed by ligating a 9193-bp ClaI fragment of pJc1/core ( TC ) and a 1381-bp ClaI fragment from pJc1/NS2 ( AP ) D88 45 ., Plasmids containing NS2 mutations in a Jc1/Gluc2A background were recently described 46 ., To facilitate cloning , the Gluc2A gene was inserted into pJc1/core ( TC ) by using common BsaBI and NotI sites ., The EcoRI/MluI fragment from this plasmid was then subcloned into the same sites of Jc1/Gluc2A plasmids containing NS2 mutations , which served to simultaneously introduce the TC tag and excise the Gluc2A gene ., Plasmids containing suppressor mutations were created in a similar manner ., To construct lentiviral vectors containing fluorescent protein-tagged cellular markers , pLenti4/EGFP ( Invitrogen ) was | Introduction, Results, Discussion, Materials and Methods | Hepatitis C virus ( HCV ) core protein is directed to the surface of lipid droplets ( LD ) , a step that is essential for infectious virus production ., However , the process by which core is recruited from LD into nascent virus particles is not well understood ., To investigate the kinetics of core trafficking , we developed methods to image functional core protein in live , virus-producing cells ., During the peak of virus assembly , core formed polarized caps on large , immotile LDs , adjacent to putative sites of assembly ., In addition , LD-independent , motile puncta of core were found to traffic along microtubules ., Importantly , core was recruited from LDs into these puncta , and interaction between the viral NS2 and NS3-4A proteins was essential for this recruitment process ., These data reveal new aspects of core trafficking and identify a novel role for viral nonstructural proteins in virus particle assembly . | Hepatitis C virus ( HCV ) infects almost 200 million people worldwide , causing both acute and chronic liver disease ., Although some antiviral treatments exist , they are not fully effective against all HCV genotypes and have serious side effects ., In order to develop more effective treatment strategies , a better understanding of how HCV replicates in infected cells is required ., In our study , we developed methods to visualize early steps in HCV particle assembly by fluorescently labeling core protein , a structural component of the virus ., Soon after protein translation , core trafficked to the surface of large , immobile lipid droplets that were adjacent to sites of virus assembly ., Core was also observed in highly motile puncta that traveled along microtubules ., By using inhibitors of virus assembly and assembly-deficient viral mutants , we showed that core is recruited from lipid droplets into these puncta , and that this process was mediated by the interaction of HCV nonstructural proteins ., Our work describes new methods to study the trafficking of core protein in infected cells , allowing us to better define aspects of infectious HCV particle assembly . | medicine, biology | null |
journal.ppat.1000408 | 2,009 | The Defective Prophage Pool of Escherichia coli O157: Prophage–Prophage Interactions Potentiate Horizontal Transfer of Virulence Determinants | Horizontal gene transfer ( HGT ) is a major mechanism involved in bacterial evolution ., In HGT between bacteria , viruses known as bacteriophages ( or phages ) play particularly important roles as gene transfer vehicles 1 , 2 ., Incoming temperate bacteriophages parasitize their hosts by integrating their genomes into the host genetic material ., The additional genetic information that they provide to the host bacterium encodes various novel abilities , such as niche adaptation and the production of new virulence factors 2 , 3 ., Although phage-mediated HGT was first described in the 1950s in the conversion of Corynebacterium diphtheriae strains that did not produce a toxin to strains that did 4 , studies in recent decades have identified a number of virulence determinants carried by phages 1–3 , 5–7 ., Furthermore , because numerous bacterial genomes have been sequenced , it has become increasingly clear that many bacterial genomes contain multiple prophages carrying a variety of genes 8 ., However , the prophages identified from the genome sequences often contain genetic defects , such as deletions or disruptions of genes required for phage induction and propagation ., Thus , such prophages are regarded simply as genetic remnants , and investigators tend to ignore the possibility that they might function as mobile genetic elements or participate in HGT ., Enterohemorrhagic Escherichia coli ( EHEC ) comprise a distinct class of E . coli strains that cause diarrhea , hemorrhagic colitis , and hemolytic uremic syndromes 9 ., Among the various EHEC strains , the most dominant are the strains of serotype O157:H7 10 ., The genome of EHEC O157:H7 strain Sakai ( referred to as O157 Sakai ) contains 18 prophages ( Sp1 to Sp18 ) and 6 prophage-like elements ( SpLE1 to SpLE6 ) , amounting to 16% of the total genome 11 , 12 ., These Sps and SpLEs have carried many virulence-related genes into the O157 Sakai genome , including the Shiga toxin genes ( stx1 and stx2 ) , a set of genes for a type III secretion system ( T3SS ) , numerous T3SS effector proteins , and transcriptional regulators for T3SS gene expression 11 , 13 ., A recent genomic comparison of O157 strains has further revealed that variation in prophage regions is a major factor generating the genomic diversity among O157 strains 14 , 15 ., An initial analysis indicated that , among the 18 Sps , 11 ( Sp3–Sp6 , Sp8–Sp12 , Sp14 and Sp15 ) retain features of lambdoid phages , one ( Sp13 ) has features similar to those of P2 , one ( Sp1 ) contains P4 features , and one ( Sp18 ) retains Mu features ., The other four Sps ( Sp1 Sp7 , Sp16 , and Sp17 ) were unable to be assigned to particular phage families due to their chimeric or highly disrupted genomic backbones ., Most of the lambdoid prophages resemble one another and contain various genetic defects ranging from frame-shift mutations to deletions and insertions of so-called insertion sequence ( IS ) elements 12 ., Thus , a functional analysis of the prophage pool of O157 Sakai could reveal whether defective prophages have any biological activity and , perhaps more importantly , whether they have the potential to disseminate virulence factors among bacteria ., In the present study , we used bioinformatic analyses to re-evaluate the genomic structures of each Sp and to define their genetic defects by comparing them with their respective well-characterized prototype phages ., We then systematically analyzed each Sp for its ability to excise itself from the host genome , replicate , and package its phage DNA ., Our results indicate that many of the apparently defective prophages can excise themselves , replicate , and be released from O157 cells as particulate DNA ., Furthermore , using Sp derivatives carrying a chloramphenicol-resistance ( CmR ) gene , we demonstrated the transferability of apparently defective prophages to other E . coli strains ., Our data indicate that defective prophages in the O157 prophage pool are not simply genetic remnants but have significant potential to act as mobile genetic elements that can mediate the spread of virulence-related genes from O157 to other bacteria ., The results further suggest that various inter-prophage interactions in the prophage pool potentiate the biological activities of the defective prophages ., The results of a genomic comparison of 18 Sps with their corresponding prototype phages are summarized below ( Figure 1 and Table 1; see Figure S1 for more detail ) ., i ) Lambdoid prophages: Among the 11 prophages with well-conserved lambdoid features , Sp3 and Sp8 lack repressor and anti-repressor functions ( CI and Cro ) , which are at the center of the regulation of lysogenization and induction of lambdoid phages 16–19 ., In the other lambdoid prophages , the repressors of Sp11 and Sp12 have been disrupted and those of Sp4 , Sp6 , Sp9 , and Sp10 lack a peptidase domain that is required for SOS-induction ( Figure 1A and Figure S2 ) ., Integrase ( Int ) , which mediates a bidirectional process of phage genome integration and excision 20 , has been disrupted in Sp3 , Sp11 , and Sp12 ( Figure 1A ) ., Although Sp4 and Sp14 appear to encode intact integrases , their putative excisionase ( Xis ) proteins that regulate the directionality of the Int function 21 lack DNA binding motifs ( data not shown ) and are probably non-functional ., Replication initiator protein O and elongation protein P 22–24 are apparently functional in all lambdoid prophages except for Sp3 ( Figure 1A ) ., Whereas six ( Sp6 , Sp8 , Sp9 , Sp11 , Sp14 and Sp15 ) have λ-type helicase loaders , three ( Sp4 , Sp10 and Sp12 ) have DnaC-type helicase loaders and Sp5 has an elongation protein almost identical to that of phage HK022 , which belongs to the DnaB family ( Figure S3 ) ., Three Sps ( Sp3 , Sp5 , and Sp15 ) have a general recombination system similar to that of phage λ , consisting of Exo , Bet , and Gam proteins ( Figure 1A ) , but those of Sp3 and Sp15 have been disrupted ., Seven Sps ( Sp4 , Sp6 , Sp9–Sp12 , and Sp14 ) possess a different type of recombination system that contains enterobacterial exodeoxyribonuclease VIII ( Exo VIII ) -type proteins instead of the λ-type exonuclease , but this system is intact only in Sp10 ., All lambdoid Sps , except for Sp4 , encode intact terminase , the key enzyme for DNA packaging 25 , 26 ., In Sp4 , the nu1 gene for the terminase small subunit protein has been disrupted by an IS insertion ( Figure 1A ) ., The morphogenesis regions of Sp3 , Sp4 , Sp8 , Sp11 , Sp14 , and Sp15 follow the gene organization of λ 18 , 26 , but Sp6 , Sp9 , Sp10 , and Sp12 exhibit slightly different gene organizations in the head formation region ( Figure 1A ) ., Of these 10 Sps , all genes for morphogenetic function are conserved only in three ( Sp8 , Sp10 , and Sp11 ) ., Sp15 ( Stx1 prophage ) also contains multiple defects in the morphogenic functions ., Most of the putative morphogenic genes of Sp5 ( Stx2 prophage ) differ from those of λ and remain uncharacterized ., However , another O157 Stx2-converting phage , called 933W , contains a set of genes nearly identical to that of Sp5 and has been shown to be fully active 27 ., All 11 lambdoid Sps encode the Q protein , a regulator of late transcription ., A full set of genes for cell lysis is also present in all lambdoid Sps , although some variation in gene organization is observed ( Figure 1A ) ., In Sp5 , an IS-insertion has occurred upstream of the lysis region , but it has not disrupted any protein-coding genes; indeed , an Sp5 derivative has been shown to transfer from O157 Sakai to K-12 28 ., Based on our in silico analysis of the 11 lambdoid Sps , we predicted that ( 1 ) three ( Sp3 , Sp11 , and Sp12 ) would no longer be able to excise themselves from the host chromosome and ( 2 ) the other eight prophages would be excisable , but seven ( Sp4 , Sp6 , Sp8–Sp10 , Sp14 , and Sp15 ) would have some defects in morphogenesis or other functions , and only Sp5 would be capable of the full complement of viral functions ( Table 1 ) ., ii ) P2 and P4-like prophages: Phage P2 functions as a helper for the satellite phage P4 29–33 ., The P2–P4 couple in O157 Sakai is atypical because homologues of the P2 ogr gene and the P4 ε gene 30 , 31 are present in Sp2 ( P4-like phage ) and Sp13 ( P2-like phage ) , respectively ( Figure 1B and 1C ) ., In addition , Sp13 lacks most phage functions , including most of the morphogenetic genes ( Figure 1B and Table 1 ) ; thus , it may no longer propagate by itself or work as a helper for Sp2 ., iii ) Mu-like prophage: Sp18 is predicted to be intact and spontaneously inducible like the prototype Mu phage because most features for Mu-like phages 34 , 35 are conserved ( Figure 1D ) ., Sp18 also contains an invertible host-specificity region , but the encoded genes are distinct from those of Mu 35 ., iv ) Others: The other four prophage genomes have been severely degraded , but all may have been derived from lambdoid phages because many of their residual genes are homologous to λ genes ( Figure 1E ) ., Reassessment of the Sp17 region revealed that two prophages ( referred to as Sp17a and Sp17b ) have been integrated in tandem in this region ., Sp7 shows interesting chimeric features of lambdoid and P4-like phages ., Sp7 encodes a P4-like Int , as well as a P4-like Xis ( Vis-homologue ) 36 ., In addition , Sp7 contains a gene similar to the P4 α gene , but the gene has been disrupted by multiple frame-shift mutations ., To experimentally evaluate the inducibility of each Sp , we first examined the amplification of prophage DNA upon mitomycin C ( MMC ) treatment of O157 Sakai cells using an oligo DNA microarray ., Cell lysis started 2 to 3 hr after the addition of MMC ( 1 µg/ml ) to the early log-phase culture , and the optical density ( OD ) returned to basal levels within 6 to 8 hr as a result of cell lysis ( Figure S4 ) ., We isolated total cellular DNA from aliquots of cultures at 1-hr intervals from 0 hr to 4 hr after the addition of MMC ., We then analyzed the total DNA using the microarray ( Figure 2A ) ., We observed selective amplification of Sp5 , Sp13 and Sp15 regions , although amplification of the Sp13 regions was delayed relative to that of the other two regions ., Interestingly , the Sp5-flanking regions exhibited significant amplification ( R1 and R2 in Figure 2A ) ., The Sp15-flanking regions also showed substantial amplification ( R3 and R4 in Figure 2A ) , but the amplification of the R3 and R4 regions was asymmetric and smaller than that for the Sp5-flanking regions ., A similar phenomenon in phage λ is known as the “regional replication” of prophage-flanking regions ., In “regional replication , ” upon induction , the chromosomal regions flanking the λ prophage genome replicate together with the λ genome that remains to be excised from the chromosome 37 ., Similar phenomena have also been reported for other lambdoid phages 38 ., Importantly , the amplified prophage-flanking regions of Sp5 and Sp15 included prophages Sp4 and Sp14 , respectively ( Figure 2A ) ., Our preliminary microarray analysis of a spontaneous Sp5 deletion mutant confirmed that the amplification of the Sp5-flanking region in response to MMC treatment requires the presence of Sp5 ( data not shown ) ., We also analyzed transcriptional changes of the prophage genes upon MMC treatment using the microarray ( Figure 2B ) ., A large number of chromosomal genes were up- or down-regulated by MMC treatment ( data not shown ) , as described in K-12 39 , 40 ., Among the prophage genes , we observed a marked increase in the transcript levels of the Sp5 , Sp13 , and Sp15 genes , especially in their early regions , which is in agreement with their selective DNA amplification upon MMC treatment ( Figure 2A ) ., No significant transcriptional changes in response to MMC treatment were detected for most genes of other prophages , except for those of Sp1 ., Although this highly degraded lambdoid prophage showed no DNA amplification upon MMC treatment ( Figure 2A ) , many of its residual early genes exhibited clear induction ., The biological significance of this phenomenon is unknown ., To determine whether the Sps amplified by MMC treatment are excised from the host chromosome into a circular form and whether other Sps are excisable but to a much lesser extent , we looked for the presence of circular forms of all Sps using PCR amplification of the attachment site ( attP ) -flanking regions that are generated by excision and circularization ( Figure 3 ) ., In MMC-treated O157 Sakai cells , we detected circularized DNA not only from Sp5 , Sp13 , and Sp15 but also from Sp6 , Sp7 , Sp9 , and Sp10 ( Figure 3B ) ., In addition , the circular forms of Sp4 and Sp14 genomes were detected , although the amount of circularized Sp14 was significantly lower than those of the other prophages ., Furthermore , we also detected the circular form of DNA for all of these prophages , except for Sp4 and Sp14 , in O157 Sakai cells that had not been treated with MMC ., These results indicate that the nine prophages can be excised into a circular form by MMC-mediated or spontaneous induction and that the Sp4 and Sp14 genomes , which were amplified by regional replication , can also be excised and cyclized ., Circularized Sp18 DNA was not detected , which is consistent with previous results for the prototype Mu phage 34 , 35 ., By sequencing the PCR products obtained in this analysis , we confirmed that the circularized prophage genomes arose by site-specific recombination between the left and right phage attachment sites ( attL and attR ) ., This analysis also allowed us to precisely determine the core attachment sequences of these nine prophages ( Figure S5 ) ., The results largely agreed with our previous predictions , except for the results with Sp9 12 ., The attL and attR sites of Sp9 are each located 121 bp upstream of the predicted positions , and we identified the true core sequence of 28 bp ., We also used the same strategy to analyze six prophage-like elements ( SpLEs ) of O157 Sakai ., However , we detected no excised and circularized DNA from any of these elements , either in untreated cells or in cells treated with MMC ( data not shown ) ., This suggests that these elements have no ( or very low ) mobility or require other types of stimuli to be mobilized ., Because we found that many Sps are excised into circular forms , we quantified the circularization and/or replication of these Sps using quantitative PCR ( qPCR ) ( Figure 3 ) ., In untreated cells , we detected similar amounts of circularized DNA of Sp5 , Sp6 , Sp7 , Sp13 , and Sp15 and a slightly lower level of circularized Sp9 DNA ., In cells treated with MMC , the relative amounts of circularized DNA of Sp5 , Sp13 , and Sp15 increased to approximately 300 , 30 , and 40 times higher than their levels in untreated cells , respectively ( Figure 3C ) ., In contrast , the levels of Sp6 , Sp9 and Sp10 were lower than in the untreated cells ., It is noteworthy that , in complete agreement with the results of the qualitative PCR analysis ( Figure 3B ) , considerable amounts of circularized Sp4 and Sp14 were generated in the MMC-treated cells , whereas hardly any was detected in the untreated cells ., Sp18 also proved to be non-inducible by MMC , as described for phage Mu 35 ., DNA microarrays were used to monitor prophage induction 38 , 41 ., Microarray analysis can provide a gross image of the replication pattern of prophages augmented by MMC treatment , as seen in Figure 2 , but it cannot detect spontaneous induction of prophages ., Thus , as our present data show ( Figure 3 , see also Figure S6 ) , qPCR analysis is required to obtain the true picture of prophage induction ., This may also be true in transcriptome analysis of prophage genes ( Figure 2B ) ., In general , the stability of a prophage is tightly coupled with the physiology of the host cell ., Under conditions that generate DNA injury—in the present study through MMC treatment—prophages are de-repressed by a RecA-mediated mechanism ( the SOS response ) to enter the lytic pathway 42 ., The RecA protein stimulates self-cleavage of the repressor protein , which leads to the expression of genes required for the lytic pathway 43 , 44 ., The non-inducible nature of Sp6 , Sp9 and Sp10 by MMC treatment is consistent with the fact that the peptidase motif is missing in the repressors of these lambdoid Sps ( Figure S2 ) ., Phage P2 is insensitive to the SOS response and is thus non-inducible by MMC treatment , because its repressor intrinsically lacks the peptidase motif 30 ., The repressor of Sp13 , a P2-like prophage , also lacks the peptidase motif ., Thus , the MMC-mediated induction of Sp13 observed in this analysis is remarkable ( Figures 2 , Figure 3 , and Figure S6 ) ., Although the mechanism is yet to be elucidated , a P4 ε-like gene encoded on Sp13 ( Figure 1 and Figure S2 ) may be involved in this unique behavior because the P4 ε gene product de-represses the P2 genome by binding to the P2 repressor 30 , 31 ., To investigate whether the Sps that were circularized and replicated by spontaneous or MMC-mediated induction could be packaged into phage particles , we first attempted field inversion gel electrophoresis ( FIGE ) analysis of DNA isolated from phage particles ., Particles were taken from supernatants of bacterial cultures that were either treated with MMC or left untreated ( Figure 4A ) ., In the untreated sample , we detected packaged DNA of Sp5 , Sp10 , and Sp18 ., However , upon MMC treatment , a large amount of Sp5 DNA accumulated and generated extensive smearing that prevented the visualization of minor species of packaged phage DNA ., We therefore quantified particulate DNA of each Sp by qPCR using the same set of PCR primers used for the quantification of intracellular phage DNA ( Figure 4B ) ., In the untreated sample , we detected DNase-resistant forms of DNA for at least five prophages ( Sp5 , Sp7 , Sp9 , Sp10 , and Sp15 ) ., They included two Sps ( Sp9 and Sp15 ) that contain genetic defects in head formation ( Figure 1 and Table 1 ) ., This result suggests that the defects of these two prophages were complemented , probably by other prophages that provided all or some of the gene products required for head formation ., The amount of Sp6 DNA was marginal compared with the control chromosomal DNA ( CB1 and CB2 in Figure 4B ) ., The Sp13 DNA appears to be inefficiently packaged or unstable ., As expected from the data on phage Mu , the Sp18 DNA was efficiently packaged ., In the MMC-treated sample , a large amount of packaged Sp5 DNA was detected , reaching levels of ≥1010 molecules per milliliter of culture ., Although at much lower levels , we detected considerable amounts of packaged DNA from at least six other Sps ( Sp4 , Sp7 , Sp10 , Sp13 , Sp15 , and Sp18 ) ., Interestingly , this group includes Sp13 , which also lacks the genes for head formation and DNA packaging ( Figure 1 and Table 1 ) ., Of the two Sps ( Sp4 and Sp14 ) whose genomes are amplified only by the regional replication of Sp5 and Sp15 , respectively , we detected packaged Sp4 genomic DNA , although it also contains defects in head formation and DNA packaging ., Thus , these defects of Sp13 and Sp4 must have also been complemented by other prophages ., To examine the transferability of packaged Sp genomes , we marked the eight Sps ( Sp4–Sp7 , Sp9 , Sp10 , Sp13 , and Sp15 ) by replacing “moron” genes of each phage genome , which are not required for phage propagation , with a CmR gene cassette ( Figures 5A and 5B ) ., The stx2 gene in Sp5 and stx1 gene in Sp15 were replaced with the cassette ., Incorporation of the CmR cassette into the prophage genomes did not affect DNA packaging because the levels of particulate phage DNA detected for each CmR-derivative were similar to those observed for the wild-type O157 Sakai ( data not shown ) ., To examine whether the CmR marker can be transferred to two K-12 derivatives ( strains MG1655 and MC1061 ) , we analyzed the culture supernatants prepared from O157 Sakai containing each CmR-marked Sp derivative with or without MMC treatment ., We found that the CmR gene on four Sps ( Sp5 , Sp6 , Sp10 , and Sp15 ) is transferable to K-12 and stably maintained , although the efficiency of transfer was low in all cases except that of Sp5 ( Table 2 ) ., Among the four Sps , three ( Sp5 , Sp6 , and Sp10 ) were integrated at the same chromosomal loci in K-12 , as in O157 Sakai ( Figure 5C ) ., The integration site of Sp10 was already occupied by the Rac prophage in K-12 , but the Sp10 derivative was integrated in tandem with Rac using the attR sequence of Rac as the attB site ( Figure 5D ) ., In contrast , the Sp15 derivative was not integrated into the yehV locus in K-12 , the chromosomal locus where Sp15 is present in O157 Sakai ( Figure 5C ) ., This suggests that recombination occurred between Sp15 derivatives and other Sps that allowed the transfer of the CmR gene ( by which the stx1 gene was replaced ) to K-12 ( see the next section ) ., Among the three Sps that were successfully transferred to K-12 , the Sp5 derivative produced infective phage particles in K-12 ( Table 2 ) ., In contrast , we could not detect the production of infective Sp6 or Sp10 derivatives in K-12 ., This result suggests that these two Sps may require the support of other Sps , available only in the O157 cell , to produce infective phage particles efficiently ., To analyze the CmR-marked Sp15 ( Sp15Δstx1::CmR ) transductants , we first performed PCR scanning analysis of the stx1-flanking region of an Sp15Δstx1::CmR-transductant of K12 MG1655 ( Figure 6A , see Figure S7 for more details ) ., The results indicated that the transductant contains an Sp15Δstx1::CmR-derived DNA segment covering the stx1 region , but also that some recombination had occurred between the P and stx1 ( replaced by the CmR cassette ) genes and between the stx1 and nu1 genes in the Sp15Δstx1::CmR genome ., By DNA sequence homology analysis between Sp15 and other Sps , we found that , although many lambdoid Sps contain one or more genomic segments that are highly homologous to the stx1-flanking region of Sp15 , only Sp5 contains both segments homologous to the upstream and downstream regions of the stx1 gene ( Figure S7 ) ., These Sp5 segments are also present in the upstream and downstream regions of the stx2 gene ., This suggested that the CmR Sp15 derivative transferred to K-12 may have been generated by recombination between Sp15Δstx1::CmR and Sp5 ., We therefore analyzed the genome of the CmR Sp15 derivative by PCR using two primer pairs: those specific to the CmR cassette and the Sp5 P gene and those specific to the CmR cassette and the Sp5 Nu1 gene ( Figure 6B and Figure S7 ) ., The two primer pairs yielded 4 . 9-kb and 6 . 4 kb amplicons , respectively , both of which were absent in the donor O157 Sakai derivative containing Sp15Δstx1::CmR ., Furthermore , we confirmed that the Sp15Δstx1::CmR transductant contains an Sp5-like phage in the wrbA locus , the integration site of Sp5 ( Figure 6C and Figure S7 ) ., All these data indicated that the CmR cassette-carrying phage is a chimeric phage that was generated by replacing the stx2 regions of Sp5 with the stx1 region of Sp15Δstx1::CmR ., We analyzed 21 additional Sp15Δstx1::CmR transductants using the same methods ., The results indicated that all of the transductants contain chimeric phages of Sp15Δstx1::CmR and Sp5 ( data not shown ) ., However , the data from our preliminary sequence analysis of the PCR products covering recombination points suggested that several ( at least four ) types of chimeric phages had been generated ( more details of these chimeric phages will be described elsewhere ) ., It may also be worth noting that this phenomenon was observed only in MMC-treated O157 cells ( Table 2 ) ., Finally , we performed an electron microscopic examination of phage particles that were present in the culture supernatants of MMC-treated and untreated O157 Sakai cells ( Figure 7 ) ., The MMC-treated sample contained numerous phage particles with a short tail attached to a head approximately 56 nm in diameter ( Figure 7A ) ., The dominant induction of Sp5 by MMC treatment ( Figure 4 ) suggests that these phage particles originated from Sp5 ., In fact , K-12 strains lysogenized by CmR-marked Sp5 produced phage particles with the identical morphology ., The morphology of Sp5 ( Figure 7A ) is highly similar to that of the previously reported Stx2 phage of O157 EDL933 27 ., We were unable to detect other phage types in the MMC-treated sample ., However , in addition to Sp5 , at least two other types of phage particles were detected in the untreated sample ., The second type had a head with a hexagonal outline approximately 49 nm in diameter , which was connected by a neck to a contractile and non-flexible tail ( the uncontracted sheath is approximately 100 nm long and the contracted one is 55 nm ) ( Figure 7B ) ., The similarity to the morphology of Mu phage particles indicates that this second type most likely originated from Sp18 ., The third type had a head with an elongated hexagonal outline ( 44 wide and 95 nm long ) and a 147 nm long flexible tail ( Figure 7C ) ., This phage probably derived from some of the lambdoid prophages , but its origin is difficult to pinpoint because lambdoid prophages other than Sp5 ( including Sp10 and Sp6 ) contain very similar morphogenetic genes ( Figure 1 ) ., We also examined the culture supernatants of K-12 strains lysogenized with CmR derivatives of Sp6 and Sp10 for the production of phage particles , but no phage particle was detected in the culture supernatants of Sp6 and Sp10 lysogens ( data not shown ) ., The results of our in silico analysis of the potential activities of 18 prophages on the O157 Sakai genome indicate that all but Sp5 contain one or more genetic defects ( Figure 1 , Figure S1 , and Table 1 ) ., This suggests that the present-day O157 prophage pool may have low potential activity as mobile genetic elements to spread virulence genes , although their mobility in evolutionary history has played an essential role in the emergence of this highly virulent E . coli lineage ., Nevertheless , our systematic experimental evaluation of the Sps revealed that many have unexpectedly high potential activity to function as mobile genetic elements ., First , nine Sps could excise themselves from the chromosome and replicate in the O157 cells in response to spontaneous or MMC-mediated induction ( Figure 3 , Figure 4 , and Table 2 ) ., They can be divided into three groups according to their induction patterns:, ( i ) spontaneously inducible ( Sp6 , Sp7 , Sp9 , Sp10 , and Sp18 ) ,, ( ii ) spontaneously inducible and further enhanced by MMC-mediated induction ( Sp5 , Sp13 , and Sp15 ) , and, ( iii ) inducible only by the regional replication of other prophages ( Sp4 by Sp5 and Sp14 by Sp15 ) ., Second , most of these Sps , except Sp6 and Sp14 , were packaged into phage particles ( Figure 4 ) , although half of them ( Sp4 , Sp9 , Sp13 , and Sp15 ) contain defects in head formation or DNA packaging ., Third , we found that the CmR gene cassette on four Sps is transferable to other E . coli strains , although we used only two K-12 derivatives as recipients ( Table 2 ) ., Three ( Sp5 , Sp6 , and Sp10 ) were transferred to K-12 and stably lysogenized in the chromosome ( Figure 5 ) ., This result indicates that Sp6 , which is also defective in head and tail formation , can be packaged , although this was not clear from the particulate DNA quantification by qPCR ( Figure 4 ) ., It is also important that these three phages carry several important virulence determinants , including the stx2 genes and multiple non-LEE effector genes ( Figure 1 and Figure S1 ) ., Fourth , the CmR gene cassette inserted into the Sp15 genome by replacing the stx1 gene was transferred to K-12 , and this transfer was achieved through the generation of a chimera between Sp15 and Sp5 ( Figure 6 and Figure S7 ) ., In addition , three types of phage with distinct morphologies were detected in the culture supernatant of O157 Sakai ( Figure 7 ) ., Two derive from Sp5 and Sp18 , respectively , but the origin of the third remains undetermined ., These results indicate that many apparently defective prophages of O157 Sakai should not be regarded as simple phage remnants , but rather as active genetic elements that can potentially mediate or assist HGT of various virulence determinants encoded on the O157 genome ., The results further suggest that various types of inter-prophage interactions occur in the O157 prophage pool , and these interactions induce the biological activities of the defective prophages ., Inter-prophage interactions that most likely occurred in the O157 prophage pool complemented various defects in morphogenetic functions by providing the proteins for phage particle formation ., Although the details of this complementation remain to be elucidated , the lambdoid phages on the O157 genome share nearly identical morphogenetic genes 11 , 12 in various combinations ( Figure 1 ) ., They therefore appear capable of supplying virion proteins compatible with those of other lambdoid phages ., In fact , we identified one type of phage particle that differs from that of Sp5 but has lambdoid features in the culture supernatant of O157 Sakai ( Figure 7C ) ., In some cases , whole virion proteins may be provided by other prophages; this may be the situation with Sp7 and Sp13 ., Both have severe defects in morphogenetic functions , but are nevertheless packaged ., Furthermore , because Sp13 is the only member of the P2-like phage family in the O157 prophage pool , this type of inter-prophage interaction may occur between very different types of bacteriophages ., In the case of Sp7 , another type of interaction may complement its defect in replication function ., This highly degraded prophage lacks most morphogenesis genes , as well as repressor and antirepressor genes ., Furthermore , the replication gene , which resembles the P4 α gene , has been disrupted into three fragments ( Figure 1E and Figure S1 ) ., Nevertheless , Sp7 is spontaneously inducible and a significant amount of circularized DNA was observed to accumulate in the O157 cells ( Figure 4 ) ., Thus , the replication of Sp7 may be mediated by the replication proteins of Sp13 ( P2-like phage ) or Sp2 ( P4-like phage ) , although we cannot exclude the possibility that some ( or all ) of the fragmented polypeptides of Sp7 may still contain some replication initiation activity ., Replication or amplification of the Sp4 and Sp14 genomes is another type of inter-prophage interaction ., Their genomic DNA can be amplified only by the regional replication of Sp5 and Sp15 , respectively ., Although both lack the genes for excisionase , integrases alone appear capable of mediating their excision from the chromosome ., Thus , the two prophage genomes amplified by regional replication are excised into a circularized form ( Figure 3 and Figure 4 ) ., More interestingly , although Sp4 does not encode an intact packaging enzyme ( terminase ) by itself ( Figure 1 and Figure S1 ) , its amplified genome was found to be packaged ( Figure 4 ) ., Most likely , this packaging was carried out by the terminase of Sp14 , because the putative cos sequence of Sp4 , which needs to be digested by terminase for packaging , is nearly identical to that of Sp14 ( Figure S8 ) ., Finally , the recombination between Sp15 and Sp5 can also be regarded as an inter-prophage interaction because it occurred in the O157 prophage pool and generated new Stx1-tranducing phages ( Figure 5 and Figure S7 ) ., This type of inter-prophage interaction can occur between other lambdoid prophages of O157 as well , because they share nearly identical sequences , which can therefore recombine 11 , 12 ., Similar recombination may also occur between the resident prophages and newly incoming phages ., In this way , high levels of excision and replication of defective prophage genomes in O157 cells may provide significan | Introduction, Results, Discussion, Materials and Methods | Bacteriophages are major genetic factors promoting horizontal gene transfer ( HGT ) between bacteria ., Their roles in dynamic bacterial genome evolution have been increasingly highlighted by the fact that many sequenced bacterial genomes contain multiple prophages carrying a wide range of genes ., Enterohemorrhagic Escherichia coli O157 is the most striking case ., A sequenced strain ( O157 Sakai ) possesses 18 prophages ( Sp1–Sp18 ) that encode numerous genes related to O157 virulence , including those for two potent cytotoxins , Shiga toxins ( Stx ) 1 and 2 ., However , most of these prophages appeared to contain multiple genetic defects ., To understand whether these defective prophages have the potential to act as mobile genetic elements to spread virulence determinants , we looked closely at the Sp1–Sp18 sequences , defined the genetic defects of each Sp , and then systematically analyzed all Sps for their biological activities ., We show that many of the defective prophages , including the Stx1 phage , are inducible and released from O157 cells as particulate DNA ., In fact , some prophages can even be transferred to other E . coli strains ., We also show that new Stx1 phages are generated by recombination between the Stx1 and Stx2 phage genomes ., The results indicate that these defective prophages are not simply genetic remnants generated in the course of O157 evolution , but rather genetic elements with a high potential for disseminating virulence-related genes and other genetic traits to other bacteria ., We speculate that recombination and various other types of inter-prophage interactions in the O157 prophage pool potentiate such activities ., Our data provide new insights into the potential activities of the defective prophages embedded in bacterial genomes and lead to the formulation of a novel concept of inter-prophage interactions in defective prophage communities . | Bacterial viruses , known as bacteriophages or phages , are major factors promoting horizontal gene transfer ( HGT ) between bacteria , and this activity has sparked new interest in light of the discovery that many sequenced bacterial genomes harbor multiple prophages carrying a wide range of genes , including those related to virulence ., However , prophages identified from genome sequences often contain various genetic defects , and they have therefore been regarded as merely genetic vestiges , with no attention paid to their potential activities as mobile genetic elements ., Enterohemorraghic Escherichia coli O157 , which harbors as many as 18 prophages , is the most striking such example ., The O157 prophages carry numerous genes related to O157 virulence , but most possess multiple genetic defects ., In this study , we analyze the functionalities of O157 prophages and report that many of the apparently defective prophages are inducible and released from the O157 cells as particulate DNA and that some can be transferred to other E . coli strains ., We should therefore regard these prophages as having high potential to disseminate virulence determinants ., Our results further suggest that their activities as mobile genetic elements are potentiated by various types of interactions among the prophages , formulating a novel concept of inter-prophage interactions in defective prophage communities . | molecular biology/bioinformatics, microbiology/microbial evolution and genomics, infectious diseases/gastrointestinal infections | null |
journal.pcbi.1005482 | 2,017 | Combination therapeutics of Nilotinib and radiation in acute lymphoblastic leukemia as an effective method against drug-resistance | The persistence of chemo-resistant leukemia-initiating cells in Philadelphia-chromosome positive ( Ph+ ) B-cell Acute Lymphoblastic Leukemia ( B-ALL ) in the bone marrow is a primary mechanism responsible for disease relapse , following treatment , which occurs in the majority of patients ., B-ALL is due , in part , to chromosomal translocations ( 9;22 ) that result in the generation of a BCR-ABL fusion protein , which fosters the transformation of immature B cells 1 ., BCR-ABL+ ( i . e . , Ph+ ) leukemia has a poor prognosis; this is particularly true when matched with deletions in Cdkn2a , the gene encoding the tumor suppressor protein ARF , which occurs frequently in B-ALL 2 , 3 ., A significant breakthrough in the treatment of Ph+ ALL as well as the treatment of chronic myeloid leukemia ( CML is associated with p210 isoform , whereas ALL is associated with p190 isoform ) was the development of the tyrosine kinase inhibitor ( TKI ) Imatinib 1 ) ., This drug , and the more potent second generation drugs Dasatinib and Nilotinib , are able to selectively inhibit the BCR-ABL mutant protein and thus significantly reduce Ph+ cell counts 2 , 4 ., While TKI therapy has long-term efficacy in the treatment of CML , most ALL patients eventually relapse following treatment with TKI due to the development of resistance 5 , 6 , 7 , 8 ., Thus a common treatment protocol for ALL patients is TKI therapy until the first remission 9 , 10 followed by stem cell transplantation ., However , since stem cell transplantation itself carries many risks to patient survival , the ability to extend the efficacy of TKI therapy in Ph+ ALL patients is of great clinical interest ., Combination therapy such as Nilotinib with inhibitors of various other pathways ( MEK , AKT , and JNK ) showed greater reduction in cell viability and lowered risk of resistance 11 ., Ionizing radiation has been used for leukemia disease in limited cases , e . g ., i ) disease involve in the central nervous system ( CNS ) , potential due to ineffective penetration of chemotherapy to CNS 12 ,, ( ii ) conditioning regimen with high doses of radiation and chemotherapy prior to stem cell transplantation for patients with high risk of relapse 13 ., Taking advantage of leukemia radiosensitivity and the benefit of low dose radiation ( LDR ) in preserving bone marrow functions , we investigated whether the combination of Nilotinib and low dose radiation will be more effective treatment for BCR-ABL+ ( i . e . , Ph+ ) leukemia over Nilotinib alone ., Furthermore , to optimize the effectiveness of this combination treatment , we developed a mathematical model , parameterized via cell viability experiments under Nilotinib treatment and radiation exposure , to predict cellular response to the combination therapy ., The optimized mathematical model predicts a synergy between LDR and TKI treatment ., We propose a combined Nilotinib dose-response function after LDR that accounts for a possible synergistic interaction between LDR and TKI treatment ., Model parameters are obtained from in vitro viability measurements in the absence of TKI ( Fig 1, ( a ) ) , with zero LDR ( Fig 1, ( b ) ) and combination of LDR and TKI for several radiation doses ( Fig 1, ( c ) ) ., The model is validated by precise prediction for the drug-dose responses and radiation-dose response to combination LDR and TKI treatment ., It is important to emphasize that our model is focused on the relevance of LDR to prevent small-molecule inhibitor drug resistance ., It does not address the efficacy of a successful radiation-drug combination treatment ., Answering the question that whether the resistance is pre-existing , selected for or evolves de novo is out the scope of the current work ., We do assume a small fraction of pre-existing Nilotinib resistant subtype , as well as possibility to transform into resistant type in the presence of the drug ., Our computational model is a coupled system of ordinary differential equations representing two populations ., A Nilotinib sensitive and a Nilotinib resistant population ., The use of ordinary differential equations is very common to describe the population dynamics and emergence of a new trait 15 , 16 Logistic terms impose limitation on growth of each population , and an additional term allows for conversion from Nilotinib sensitive to Nilotinib resistant phenotype in the presence of non-zero concentration of the drug ., More specifically , we assumed the following for the dynamics of the two subpopulations of the sensitive and resistant phenotypes: The dynamical model that describes the above mechanisms is detailed in the Supplementary Information ( SI ) ., We refer to this model as the proliferation-mutation model ., To identify the response of Ph+ ALL cells to Nilotinib treatment , radiation , and the combination of both therapies , we proposed a simple functional form of this dependence and evaluated the fit against a large set of dose response data from experiment ., This simple linear dose-response model can be written as follows:, ( proliferation rate ) = r 0 - ( Nilotinib dose ) × r 1 + radiation dose × r 2 , ( apoptosis rate ) = d 0 + ( Nilotinib dose ) × d 1 + radiation dose × d 2 , ( 1 ), where the growth rate coefficients ri and di are constants that need to be identified based on the experimental results ., It is common to model dose response functions using a Hill function structure ( 3- or 4-parameter logistic function ) ., The Hill function imposes a low level of drug efficacy at low doses and a saturation of drug efficacy at high doses ., A linear response is most accurate to approximate a Hill function dose-response near IC50 does—which is the case here ., In the SI section , we show that how the above linear dose-response functions can be derived from a more common Hill -function ., Eq ( 1 ) can be rewritten as, r S = r S , 0 − ( r S , 1 + r S , 2 D ) c , r R = r R , 0 − ( r R , 1 + r R , 2 D ) c , d S = d S , 0 + ( d S , 1 + d S , 2 D ) c , d R = d R , 0 + ( d R , 1 + d R , 2 D ) c , ( 2 ), where c and D represent the Nilotinib and radiation doses respectively ., The constants rS , 0 and rR , 0 denote the proliferation rates of sensitive and resistant populations in the absence of therapy , and dS , 0 and dR , 0 denote the death rates of sensitive and resistant populations in the absence of therapy ., The coefficients rS , 1 , rR , 1 represent the dose-response relationship between Nilotinib and proliferation rate of sensitive and resistant cells , respectively ., Similarly , the coefficients dS , 1 , dR , 1 describe how Nilotinib impacts the death rate of sensitive and resistant cells , respectively ., Lastly , the coefficients rS , 2 , rR , 2 , dS , 2 , dR , 2 determine the strength of the radiation-drug interaction on proliferation and death rates sensitive and resistant cells , respectively ., These coefficients were fit to experimental data sets studying proliferation and death rates at a variety of Nilotinib and radiation doses ., For example , non-negligible positive fitted values of rS/R , 2 and dS/R , 2 reveal synergistic interaction between the therapies ., We also incorporated an immediate cell-kill term after each radiation dose in accordance with the standard linear- quadratic model 17 , 18 , 19 , 20 ., According to the LQ model the effects of radiation cell kill is given by the survival fraction after the radiation exposure, Surviving\xa0fraction=exp − α × ( radiation\xa0dose ) − β × ( radiation\xa0dose ) 2 ) ( 3 ), where α and β are the radio-sensitivity parameters to be determined from the experimental data ., In short , α represents the rate of cell kill due to single tracks of radiation and β represents cell kill due to two independent radiation tracks ., The linear quadratic model is widely used due to its excellent agreement with empirical data for a wide range of radiation doses ., In SI section we explain how to incorporate the linear quadratic framework for surviving fraction into our mutation-proliferation model ., The above mathematical framework can predict the population fraction ( cell viability ) at each day given the initial values of Nilotinib and irradiation doses ., We fit the model parameters in an iterative fashion ., All parameter estimation are obtained by finding optimal parameter set that minimizes the square root distance of solution of Eq ., S3 ( in SI ) with respect to the time series viabilities at days 0 , 3 , 6 , 8 , 10 ., The steps are as follows:, We first measured the time-dependent cell viability of Ph+ ALL cells in vitro in response to Nilotinib , radiation , and combination therapy with both Nilotinib and radiation ( Fig 3, ( a ) ) ., For the radiation-only arm , we observed an initial large reduction inviability by day 6 for 2 Gy and day 8 for 4 Gy ., Experiments using Nilotinib monotherapy showed an incremental reduction in cell viability over the first 6 days to about 57 . 2% ., Subsequently the cell populations began to develop a resistance to the drug and viabilities began to increase ., When used in combination , Nilotinib + radiation induced a more effective initial cell killing and the cancer cell population was controlled at very low numbers ( under 10% viability ) for the duration of the experiment ., In other words , there was nearly no cell population recovery , and resistance to Nilotinib was not developed ., Under the Nilotinib + 4 Gy treatment arm , the cell population was entirely eliminated ., Thus , the combination therapies were able to not only elicit a larger reduction in cell population , but also to maintain control of low cell viabilities without the development of resistance over a longer time period ., The use of Triciribine was able to keep the cell viability as low as Nilotinib+ radiation group ( Fig 3, ( b ) ) ., The radiation dose-response assay revealed an LD50 of 2 Gy in the absence of Nilotinib , and 0 . 5 Gy in the presence of 18 nM Nilotinib ., The Nilotinib dose-response assay revealed an IC50 oof 18nM in the absence of radiation ., To investigate how the combined therapy provides the synergistic effect , we evaluated one of the key pathways associated with cell proliferation , and chemoresistance ., At day 3 after treatment , p-AKT was slightly increased by adding either Nilotinib or radiation ( Fig 4, ( a ) ) ., However , the expression level was gradually decreased at large dose of radiation ( 6 Gy ) with or without Nilotinib ( Fig 4, ( b ) ) ., Notably , the combined therapy eventually ( day 7 and10 ) decreased the p-AKT even at 2 Gy ( Fig 4, ( c ) ) ., Using the numerical solutions of the proliferation-mutation model optimized for the best fit with experimental data points , we determined the coefficients in the Nilotinib dose-response function , ri and di , as well as the transformation rate ( Table 1 ) ., The optimized coefficients indicate an increase in fitness of resistant cells relative to that of sensitive cells in the presence of Nilotinib ., ( Fitness is defined as the difference between division rate and apoptosis rate in the cell population . ), Using a numerical sensitivity analysis , we confirmed that these parameter estimates are robust to perturbations in the initial frequency of resistant cells in the population ., Since the experiments demonstrated an initially strong sensitivity of the populations to Nilotinib treatment , we set the frequency of resistant cells to be small ., We set this to be 0 . 1% of the total population for the remainder of investigations ., The optimized parameters are reported in Table 2 with the carrying capacity set to K = 4 . 2 in all cases ., These results indicate the presence of synergistic effects between Nilotinib and radiation ., In particular , the radiation tyrosine-kinase inhibitor interaction was strongest in the resistant population ., This is not surprising as in the presence of Nilotinib the sensitive population is already highly disadvantaged in terms of growth ., The value of transformation rate ν in this case is independent of radiation dose and is higher than the respective transformation rate in the Nilotinib-only case suggesting that radiation exposure contributes to the production of new resistant cells ., The results for population fractions ( cell viabilities ) as a function of time ( days ) were plotted in Fig 5 ., The agreement between mathematical model predictions and the in vitro measurements suggests that the proposed radiation-drug interaction term in the linear dose-response is a plausible choice ., All the above results of the parameter estimation process are summarized in the Tables 1 and 2 ., We next tested predictions of the fitted model with an independent set of experimental results ., In particular , we compared model predictions to the experimental results of the dose-response assays for Nilotinib ( in absence of radiation ) and radiation ( at 0 and 18 nM Nilotinib ) ., See Fig 6 for these comparisons ., As can be seen , the results are in very good agreement ., We then used the model to predict the Nilotinib dose-response curve with 1 , 2 , 3 and 4 Gy radiation ., The results are plotted for 0-22nM of Nilotinib ., Note that at higher doses we see a discrepancy with the experiment ( 25nM 0Gy ) ., This is due to the fact that the linear approximation of the Hill dose-response curve only applies to the initial section of the curve and is expected to break down at higher doses ., The combination therapy above was shown to reduce tumor cell viability , and through computational techniques an optimal schedule of dosing may be approached ., Given the potential toxicity of Nilotinib to normal tissues , we first assumed a standard , constant dose of 18 nM Nilotinib ., Next , we considered a five-day radiation treatment protocol , where the summed dose of the protocol is constant ., As an example , we use a total dose of 2Gy; which , as can be seen in Fig 5 , has room for improvement with the combination therapies and does not irradiate the cells to a viability that is not interesting mathematically ( 4Gy in Fig 5 , for example ) ., The aforementioned result will serve as a comparison for our proposed protocol , which will attempt to minimize the total tumor cell viability at day 10 ., The control parameters are the radiation doses given at days 0 , 1 , 2 , 3 and 4 denoted by Di ( i = 1 , 2 , 3 , 4 , 5 ) ., Placed under the constraint of the total allowable dose , the control parameters were varied by a nonlinear constrained minimization protocol that sought to minimize the tumor cell viability at day 10 ., Under this minimization protocol , the minimum tumor cell viability was determined by searching the potential dose regimen in the answer space and producing the dosing protocol that best minimized the tumor cell viability at 10 days ., This resulted in an optimal dosing schedule of ( in Gy ) D1 = 0 . 9371 , D2 = 0 . 5139 , D3 = 0 . 6445 , D4 = 0 . 045 , D4 = 0 . 0064 , which front-loads the radiation in the first three days ., Notice that the negligible value for D4 indicates that optimal solutions are effectively that of a 4-fraction protocol , with variable doses per fraction ., In between any two radiation doses , the cells undergo repopulation ., For acute radiation protocols , the linear dose response function predicts a change in proliferation potentials of both sensitive and resistant cells which depends on the total radiation dose at day 0 ., For a fractionated protocol , we assume the the proliferation potentials of the two cell types between the kth and k+1th radiation doses depends on total radiation dose until fraction k ., The model prediction for total cell viabilities versus time for the optimal fractionation protocol is plotted in Fig 7 and compared with a constant 5-fraction radiation protocol ( Di = 0 . 4 ) and 2Gy acute radiation treatment ( D1 = 2Gy , D2 = D3 = D4 = D4 = 0 Gy ) ., At larger time , that is , greater than 10 days , the optimal protocol has the lower cell viability more importantly in a downward trend beyond the 10 day point , whereas the acute treatment begins an upward trend and the constant fractionation protocol does not suppress the cell viability to the levels of the optimal or acute protocols ., This suggests the synergistic interactions between the therapies is heightened with larger initial doses of fractionated radiation , and the optimal protocol for long term suppression falls between acute dosing and constant fraction protocols , which results in the front-loading protocol ., This dose dependent fractionation may be generalized to various scenarios in the clinical setting as determined by clinical status ., Further , the computational model may be extended to consider alternative Nilotinib dosing strategies in combination of the radiation protocol to minimize the cell viability at day 10 ., Concurrent use of the 5-day optimal fractionation protocol with a varied Nilotinib dosing strategy shows an alternative , though equally efficacious ( similar cell viability at 10 day ) strategy ., This alternative Nilotinib dosing strategy maintains the average daily dose of 18 nM , however , it begins at lower concentration and increases throughout the course of treatment ., Specifically , the Nilotinib dose over the first 3 days is 10 nM , followed by 18 nM for 4 days , and finally 26 nM for 3 days , essentially back-loading the Nilotinib onto the irradiated cells ., After 10 days , the dose returns to 18 nM daily ., This protocol is visualized in Fig 7 as well , where at larger times the new strategy of Nilotinib dosing reduces the cell viability as well as the previous protocol ., This is clinically relevant as dose titration is common , if needed , and there is no trade off with treatment efficacy should this dosing schedule be indicated ., Although the clinical outcome of BCR-ABL leukemia has been improved with the advent of TKI therapy , overcoming chemo-resistance has been a major challenge ., This study demonstrated that low dose irradiation combined with Nilotinib provided enhanced and prolonged efficacy for leukemic cells in vitro ., A companion theoretical model provided good agreement with experimental results with an opportunity for further optimization to enhance treatment efficacy ., We explored here the possibility of using low dose radiation as a first line therapy in combination with chemotherapy to enhance the effectiveness controlling BCR-ABL leukemia ., There are mainly two reasons for such a combination therapy approach:, a ) advent of image guided targeted radiation allow more focused dose delivery 21 , 22 , 23, b ) known radiosensitivity of leukemia but not known if low dose radiation with chemotherapy could be an effective alternative to control over the TKI drug resistance ., We found that combining low dose of radiation with a commonly used TKI drug could not only substantially reduce cell population , but also maintain low cell viabilities without the development of resistance over a longer time period ., To elucidate the mechanism for which low dose radiation provide beneficial effect , we investigated the role of AKT pathway using western blot analysis ( Fig 4 ) ., Our results revealed that high dose radiation ( i . e . 6 Gy ) reduced the phosphorylation of AKT ., Interestingly , even low dose radiation was adequate to inhibit the phosphorylation of AKT when Nilotinib was also used ., To support of this , the cell viability of chemo-resistant cells ( with increased p-AKT ) was significantly reduced when AKT inhibitor ( Triciribine ) was used in combination with Nilotinib ., These results suggest that radiation can play roles not only in a conditioning regimen before stem cell transplant but also an alternative for an AKT inhibitor ., In other words , low dose radiation therapy could be an alternative treatment option for AKT inhibitor in ALL patients ., To our best knowledge , this is the first report that demonstrates the promising combined efficacy of Nilotinib and radiation for ALL , with minimum damage to vital organs ., Because of low dose with limited toxicity , it can be delivered to the whole body ., Since our study was performed in normal cell culture , however , in vivo study is required to take into consideration of microenvironmental factors ., We also constructed a dynamical model to explain the observations and to predict response to additional combination therapy schedules ., To tease apart the responses of Ph+ ALL cells to Nilotinib , radiation , and the combination of both therapies , we proposed a simple functional form of the combination dose-response relationship and evaluated the fit against a large set of dose response data ., In particular , we proposed a simple Nilotinib dose-response function in which radiation dose may alter the strength of the Nilotinib response in a dose-dependent fashion ., Parameter-fitting revealed an optimal parameter set that showed very good agreement with experimental results ., Analysis of this optimal parameter set revealed dose-dependent synergistic effects between Nilotinib and radiation response ., In particular the radiation tyrosine-kinase inhibitor interaction is strongest in the resistant subpopulation of cells ., Our analysis also demonstrated that the model predictions are robust to variation in the initial frequency of resistant cells ., We compared fitted model predictions with an independently generated second set of experimental data and found good agreement ., We next utilized the validated model to investigate optimal combination strategies for Nilotinib and radiation in Ph+ ALL ., As a simple test , we assumed a standard 18 nM Nilotinib dose and investigated strategies allowing up to 2 Gy over the course of a 5-day treatment ., We determined that the optimal therapeutic schedule , given these constraints , spread most of the radiation dose over the first three days ., Thus , a ‘sweetspot’ exists between acute radiation protocols and constant treatment protocols ., These results suggest a promising direction for investigation of new treatment strategies in Ph+ ALL and for providing an optimal treatment regimen ., However , we note that all experiments ( and thus parametrization of the model ) was done in vitro ., Further in vivo studies are needed to determine treatment schedules for the clinic with more accuracy ., To summarize , augmentation of LDR-Nilotinib therapy may be beneficial to control Ph+ve leukemia resistance and the model can determine optimal dosing schedule to enhance the effectiveness of the combination therapy . | Introduction, Materials and methods, Results, Discussion | Philadelphia chromosome-positive ( Ph+ ) acute lymphoblastic leukemia ( ALL ) is characterized by a very poor prognosis and a high likelihood of acquired chemo-resistance ., Although tyrosine kinase inhibitor ( TKI ) therapy has improved clinical outcome , most ALL patients relapse following treatment with TKI due to the development of resistance ., We developed an in vitro model of Nilotinib-resistant Ph+ leukemia cells to investigate whether low dose radiation ( LDR ) in combination with TKI therapy overcome chemo-resistance ., Additionally , we developed a mathematical model , parameterized by cell viability experiments under Nilotinib treatment and LDR , to explain the cellular response to combination therapy ., The addition of LDR significantly reduced drug resistance both in vitro and in computational model ., Decreased expression level of phosphorylated AKT suggests that the combination treatment plays an important role in overcoming resistance through the AKT pathway ., Model-predicted cellular responses to the combined therapy provide good agreement with experimental results ., Augmentation of LDR and Nilotinib therapy seems to be beneficial to control Ph+ leukemia resistance and the quantitative model can determine optimal dosing schedule to enhance the effectiveness of the combination therapy . | High likelihood of evolution of resistance to therapy is common in most forms of leukemia ., This issue persists for tyrosine kinase inhibitor drug treatments as well as other forms of therapies ., In the current work , we suggest a combination therapy where Ph+ acute lymphoblastic leukemic cells are treated with low-dose radiation before chemotherapy ( Nilotinib ) ., Our in vitro results of the combined therapy accompanied with a mathematical model shows successful suppression of resistance to Nilotinib ., The mathematical model shows a synergistic interaction between Nilotinib and low dose radiation in the chemo dose response function ., Beside acute radiation we investigate low dose fractionated therapies with model predicted optimal dosing schedules . | cell death, leukemias, medicine and health sciences, radiation therapy, dose prediction methods, cancer treatment, clinical oncology, cell processes, cancers and neoplasms, mathematical models, oncology, hematologic cancers and related disorders, clinical medicine, pharmaceutics, research and analysis methods, lymphoblastic leukemia, mathematical and statistical techniques, hematology, acute lymphoblastic leukemia, cell biology, biology and life sciences, drug therapy | null |
journal.pcbi.1006275 | 2,018 | Information-theoretic analysis of realistic odor plumes: What cues are useful for determining location? | Diverse species throughout the animal kingdom use olfactory cues for navigation tasks critical to survival , including locating food sources and mating partners ., However , olfactory navigation is not simple: odorants are often volatile and carried on rapidly changing currents , resulting in spatiotemporal distributions that are turbulent thereby defeating simple strategies such as gradient detection ., Consequently , recent efforts at understanding olfactory navigation have focused on identifying the viable computational strategies for making navigation decisions 1 , 2 ., Here we focus on the most basic aspect of this process: how odor samples are encoded in the first place ., Since sensory resources are finite , tradeoffs are inevitable ., For example , resources may be allocated to encoding individual samples of odor concentration at a fine level of detail , or alternatively , to encoding multiple samples , either in space or in time , but at a coarser resolution for concentration ., In this study , we investigate the implications of these and related tradeoffs , using the tools of information theory ., Specifically , we compare an array of sampling and encoding strategies , asking to what extent they capture information about location within an olfactory environment ., There are several aspects of the statistics of an odor plume that can give clues as to the location of the source 3–7 ., For example , the mean concentration varies smoothly in lateral and longitudinal directions ., However , animals do not base their navigation decisions on mean concentration , as the time it takes to obtain stable estimates of mean concentration exceeds the typical time taken by animals to make navigation decisions 8–10 ., Other olfactory features that have been proposed as useful for navigation decisions include the time between odor encounters 11–13 and intermittency ( the probability of the odor concentration above threshold ) 4 ., However , as for mean concentration , obtaining stable estimates of these quantities takes more time than animals typically use for navigation decisions ., Hence averaged quantities—even if aided by other sensory inputs—are probably not used to guide navigation decisions ., These considerations motivate our focus on what can be learned from brief , localized samples ., We do not address the issue of how to integrate odor samples with other sources of information ., A key starting point for our analysis is the explicit recognition that the resources available for sampling and encoding an odor environment are finite , and that it is natural to quantify these resources in terms of bits ., This leads to the framework of information theory , which has the advantage that it minimizes the assumptions about the odor distribution ., As mentioned above , the sampling strategies we consider explore tradeoffs between the number of bits allocated to resolving concentration , and to sampling in space and time ., The focus on these tradeoffs is motivated by the diversity of the sampling strategies that animals use ., With regard to spatial aspects , most animals have two spatially separated antennae or nostrils which sample the olfactory environment , but the sensor spacing ranges from less than a mm to several cm ., With regard to temporal aspects , insects’ olfactory receptors are continuously exposed to odorants , while rodents take periodic samples and adjust their sniff rate based on previous measurements 14–16 ., In this article , we discuss sampling strategies based on local cues in light of how much information they provide about sampling location ., To compare different sampling strategies , we computed the information that they conveyed about location , for three realistic olfactory environments ., In each environment , odor concentration was empirically determined via physical measurements , planar laser-induced fluorescence 17 ., We chose to use physical measurements of actual plumes not only to avoid the assumptions made by models of turbulence or the complexities of numerical simulations , but also because the non-idealities of physical measurements take into account the real-world issues that confront the olfactory navigator ., Although the three environments differed with regard to flow rate , turbulence , and proximity to a boundary , a number of commonalities emerged ., First , precise measurement of odor concentration is generally not useful ., That is , after allocating one or two bits to a coarse representation of odor concentration , more information about location is gained by using additional bits for encoding concentrations at nearby locations in space or time , than by using these bits to refine the representation of concentration ., We also demonstrate that using “histogram equalization” as a strategy to discretize odor concentration—which is optimal to convey information about intensity per se 18—is not optimal when the goal is to determine location ., That is , the optimal strategy for low-level sensory encoding depends on the ultimate use of the information ., Finally , with regard to sampling in time , we find that the additional information gained from multiple samples is preserved even if the temporal order of the samples is ignored , and this provides a rationale for simple post-receptoral processing strategies ., Odor plume data were obtained experimentally using a surrogate odor ( acetone ) released in a turbulent flow within a benchtop low-speed wind tunnel ., We imaged the odor structure using planar laser-induced fluorescence ( PLIF ) ; images were subsequently post-processed into calibrated matrices of normalized concentrations ., We acquired three separate datasets varying in mean flow rates and proximity to a boundary ., The wind tunnel has a test section measuring 1 m long , by 0 . 3 m tall , by 0 . 3 m wide ., We collected odor plume data at flow speeds of 5 cm/s and 10 cm/s ., Ambient air enters the tunnel through a contraction section and passes through a turbulence grid consisting of 6 . 4 mm diameter rods with a 25 . 5 mm mesh spacing ., Air exits the test section through a 15 cm long honeycomb section used to isolate the test section from a fan located in the downstream contraction ., The odor surrogate was released isokinetically through a 9 . 5 mm diameter tube on the tunnel centerline ., The tube orifice was located 10 cm downstream of the turbulence grid ., For one dataset , named boundary flow , a false floor spanning the length and width of the test section was placed directly below the release tube ., Acetone vapor was used as a fluorescent odor surrogate ., We generated the acetone vapor by bubbling a carrier gas through liquid acetone ., Because acetone is denser than air , the carrier gas consisted of a mixture of air ( 59% v/v ) and helium ( 41% v/v ) such that the odor surrogate mixture was neutrally buoyant in the wind tunnel ., We used a water bath to maintain the temperature of the odor mixture at ambient tunnel conditions ., A 1 mm thick light sheet from a Nd:YAG 266 nm pulsed laser illuminated the odor plume in the test section , causing acetone vapor in the odorant mixture to fluoresce with an intensity proportional to its concentration ., The laser sheet enters and exits the tunnel through longitudinal slits along the sides of the test section ., Plume fluorescence was imaged through a glass window in the tunnel using a high quantum efficiency sCMOS camera , with a bit depth of 16 bit , at a framerate of 15 Hz synchronized with the laser pulses ., To enhance signal-to-noise , images were binned to ( 512x512 ) pixels corresponding to a spatial resolution of 0 . 74 mm/pixel ., Raw images were processed to correct for background according to the equation, c ( t , x , y ) =1acI ( t , x , y ) F ( x , y ) , ( 1 ), where c is the normalized concentration , I is the image from the camera ( with background signal subtracted ) and F is the flatfield image ( also with the background signal subtracted ) ., The calibration coefficient , ac , was used to normalize the concentrations based on the source concentration at the tube exit ., Three datasets were collected , which had different combinations of wind tunnel flow rates and false floor configurations ( Table 1 ) ., The first condition , named fast flow , had a mean free stream velocity of 10 cm/s , and the odor mixture was released into the center of the tunnel without a false floor ., The second condition , named slow flow , had a free stream velocity of 5 cm/s , and acetone was also released into the center of the tunnel without a false floor ., The third condition ( boundary flow ) had a free stream velocity of 10 cm/s , but in contrast to the first condition , acetone was released with the false floor in place ., All datasets were collected in segments of 4 minutes ., We had a total of 40 minutes ( 36000 frames ) for the first and third condition , and 36 minutes ( 32400 frames ) for the second dataset ., The matrices of normalized concentrations provide a natural coordinate system ., Time-averaged odor concentrations and two typical snapshots for the three conditions are shown in Fig 1 ., To compare olfactory cues across different flow conditions , we chose two grids of 16 locations in each olfactory landscape ( G narrow and G wide ) ., Coordinates of the locations for the grid choices ( inlet location at the origin ) are:, G narrow = { ( x , y ) | x = ( 2 . 2 , 5 . 9 , 9 . 6 , 13 . 3 ) cm , y = ( - 4 . 4 , - 1 . 5 , 1 . 5 , 4 . 4 ) cm } , G wide = { ( x , y ) | x = ( 5 . 6 , 11 . 1 , 16 . 7 , 22 . 2 ) cm , y = ( - 2 . 6 , - 1 . 1 , 1 . 1 , 2 . 6 ) cm } ., ( 2 ), The two grids were chosen to capture the environment close to the source and further away from it above and below the centerline ., The locations are indicated as blue circles ( G narrow ) and green triangles ( G wide ) in Fig 1 ., The distances between gridpoints and the odor source are directly relevant to walking flies and other small insects ., Probability distributions of the odor concentrations of the upper half of all grid points are shown in S1 Fig . Our primary goal is to quantify the extent to which a small number of samples of odor concentration within a plume provide information about the location of the sample ., A principled approach is to use Shannon’s mutual information ( MI ) 19 for this purpose ., That is , using entropy as a measure of uncertainty , we will determine the extent to which a given encoding scheme reduces the uncertainty about the location of the sample ., Thus , our two variables of interest are location ( L ) and discretized odor samples ( M ) ; these are related in a complex statistical fashion ., Specifically , this analysis quantifies the ability to discriminate between the 16 locations of either G narrow or G wide when the only available information comes from odor intensity samples ., The choice of 16 locations per grid is somewhat arbitrary , however , in order to get stable information estimates with a given amount of data one trades off the number of locations with the number of bits using for odor coding ., We settled on 16 locations as they capture a good proportion of the environment while allowing for the analysis of coding of odor samples with up to 10 bits ., As is well-known , the MI between two random variables L and M is 19 , 20:, I ( L , M ) = H ( L ) - ∑ m ∈ M p ( m ) H ( L | m ) , ( 3 ), where H ( L ) is the ( unconditional ) entropy of L , and H ( L | m ) is the entropy of the distribution of L conditional on m ∈ M . In our context , L is the set of sampling locations G narrow or G wide and m ∈ M is a measurement of the normalized odor concentration c ( t , x , y ) ., The specific representation of c as a ( coarser ) measurement m is an integral part of the encoding schemes we consider ., We assume that the a priori probability of the locations l ∈ L are equal ., It follows that the unconditional entropy is, H ( L ) = - ∑ l ∈ L p ( l ) log 2p ( l ) = log ( | L | ) , ( 4 ), where | L | is the number of sampling locations ., Note that the MI ( Eq 3 ) is a property of the grid as a whole , not the individual points ., Since all | L | grid points have the same a priori probability , the upper bound of the MI is log 2 ( | L | ) ., If the navigator has log 2 ( | L | ) bits of information then it knows its location on the grid unambigously ., Posterior ( conditional ) distributions p ( l|m ) were calculated by Bayes theorem ., Specifically , we binned the odor concentrations c at each location p ( m|l ) and then normalized the likelihoods by p ( m ) ., The entropy of these conditional distributions are given by, H ( L | m ) = - ∑ l ∈ L p ( l | m ) log 2 p ( l | m ) ., ( 5 ), This quantity , weighted by the probability that sample m occurs p ( m ) , is summed over all m ∈ M to determine the average conditional entropy in Eq ( 3 ) ., We used two contrasting strategies for representing the odor concentration as discrete symbols ( bins ) ., In the first strategy , we divided the data into equal quantiles , i . e . we chose boundaries such that the distribution p ( m ) is uniform ., This histogram-equalization procedure maximizes the information conveyed about the odor concentration ( i . e . , M ) 20 , chap . 2 , but does not necessarily maximize the information conveyed about sampling location ., In the second strategy , we adjusted these bin boundaries to increase the amount of information about location ., Because finding the bin boundaries that yield an absolute maximum is a multidimensional discrete optimization problem , we used the following “greedy” iterative strategy to find an approximate maximum ., The first bin boundary is chosen to maximize I ( L , M ) , and is identified by an exhaustive search of the range of concentrations ., Then , iteratively , the k-th boundary is chosen to maximize I ( L , M ) while keeping the k − 1 bin boundaries fixed ., This is also a one-dimensional search over the range of concentrations , and leads to a binary subdivision of one of the bins determined at the previous step ., For analyses in which the odor at multiple temporal or spatial samples is encoded , we used the bin boundaries determined from these single-sample optimizations ., The encoding strategies we considered are specified not only by the way that each sample is encoded ( i . e . , the bin boundaries ) , but also by the number of spatial samples rspat and the number of temporal samples rtemp ., Specifically ,, S ( n bits ; r spat , r temp ) , ( 6 ), denotes an encoding strategy that uses nbits to discretize odor intensity , applies this discretization to rspat samples at nearby locations obtained at rtemp points in time ., Note that the number of bins used to discretize odor concentration is given by 2 n bits ., When investigating strategies with two sensors ( rspat = 2 ) , we take two samples at a distance of 0 . 3 cm ( four pixels ) centered around the locations specified in Eq ( 2 ) ., For sampling strategies specified by the notation of Eq ( 6 ) , bin boundaries are obtained by histogram equalization ., To indicate that the “greedy” strategy has been used for obtaining bin boundaries , we use the symbol n bits * ., The total number of bits used for encoding a sample m is given by nbits ⋅ rspat ⋅ rtemp ( or n bits * · r spat · r temp ) ., To ensure that our results do not reflect the idiosyncrasies of odor concentrations at specific locations , all calculations were repeated after jittering the grid location ., Specifically , the grid was rigidly moved from its standard location ( as given in Eq ( 2 ) ) by 0 . 74–2 . 22 mm ( 1-3 pixels ) in x and y directions , yielding a total of 49 placements ., In all figures of the results section , mutual information at these jittered locations are shown as shaded blue and green regions ., We considered encoding schemes that probed the three basic ways in which resources could be allocated to encoding the odor measurements: for resolving concentration , for sampling across space , and for sampling across time ., Here and in the other analyses below , parallel calculations were carried out for three odor environments: fast flow ( A ) , slow flow ( B ) and boundary flow ( C ) , and for two sets of locations ( narrow grid ( blue ) and wide grid ( green ) ) within each environment ., The fast flow and boundary flow conditions have the fastest inlet flow of 10 cm/s , but the boundary flow dataset was taken near a boundary where the odor surrogate’s dynamics are affected by boundary layer effects ., Hence , boundary flow is the condition were diffusion has the biggest impact; see Methods for details ., As a consequence of the more diffusive regime of the boundary flow condition the mutual information values we obtained for this condition are somewhat higher than in the other two conditions ., The slow flow dataset has an inlet velocity of 5 cm/s ., Except as noted , the analyses with different datasets and different grid choices yielded similar results ., Fig 3A1–3C1 shows results for strategies that devote all bits to encoding concentration at one point in space and time ( S ( n bits ; 1 , 1 ) ) ., As the resolution for odor concentration increases , so does MI , but only up to a point: once four bits are used to resolving odor concentration , additional resolution yields only minimal increases in MI ., When measurements are made at two sensor locations ( transversely separated by 0 . 3 cm ) , using additional bits for coding allows MI to increase beyond the plateau encountered with a single sensor ( Fig 3A2–3C2 ) ., The benefit of spatial sampling is not merely the result of having two independent samples ., Specifically , MI computed after ignoring which sample corresponded to which sensor was smaller , by up to 0 . 1 to 0 . 2 bits ( dashed curves in Fig 3A2–3C2 ) , than the MI conveyed by a coding scheme that keeps track of which sample is which ., This indicates that sampling with two sensors enables extraction of a spatial feature of the odor plume that varies along the vertical axis ., This trend is also true for different spacing between two sensors , as shown for half intersensor distance and double intersensor distance in S5 Fig . Note that in the boundary flow condition , the curves continue to increase rapidly at the limits of measurement , suggesting that MI is not close to saturation ., Encoding odor measurements at two consecutive times ( separated by 1 . 6 s ) also increases MI beyond the plateau of a single sample , but not by as much as for two spatial samples ( Fig 3A3–3C3 ) ., While each additional bit used for resolving the concentration of two consecutive samples provides greater MI , the increases become progressively less , suggesting that MI has reached a plateau when five bits of resolution are devoted to two samples separated in time ., Virtually identical results are obtained for longer intervals between samples; this is expected since MI reaches an asymptotic value as a function of sampling interval ( see section temporal encoding strategies below ) ., In the above analysis , we discretized the odor concentration into sub-intervals of equal probability , as this histogram-equalization procedure provides the greatest amount of information about the odor concentration itself 18 , 20 ., However , this does not yield the maximal MI about location , so we carried out a further analysis that explored the discretization strategy ., For the simple case of discretization into two levels , we show how the MI depends on the binarization threshold in Fig 4 ., For the boundary flow condition ( C ) the information curves are flat over a large range for the narrow grid , and has a maximum above the median for the wide grid ., For the fast flow ( A ) and slow flow ( B ) condition the maximum of information is obtained when the threshold is above the median for both grids ., This suggests the most informative samples about location occur at high concentration ., A threshold above the median exploits this feature of the odor statistics and allows better discriminability between locations ., A comparison between the bin boundaries obtained by histogram-equalization and the optimal bin boundary when binarizing odor can be seen in S2 Fig . It is evident that the optimal bin boundary occurs at a higher concentration than the median for all but the narrow grid of the most diffusive condition ., To investigate how a different choice of bin boundaries affects the results of Fig 3 , we implemented a “greedy” partitioning scheme ( see Methods ) in which the first cutpoint was chosen to yield the maximal MI about location , and then successive cutpoints were chosen so that each maximized the MI about location , given the previous partitioning ., Results ( see S3 Fig ) were very similar to the above analysis based on histogram-equalized bins ( Fig 3 ) ., Although one- and two-bit encoding schemes ( two to four partitions ) yielded more MI than histogram equalization , the plateau seen in row 1 of Fig 3 was essentially unchanged ., The advantage of encoding schemes based on two spatial or two temporal samples persisted ., The above findings show that overall , there is surprisingly little benefit to allocating coding bits to resolving odor concentration , compared to allocating them to capture several samples across space or time ., We hypothesized that resolution of odor concentration might become more important in regimes that were more diffusive , especially when coupled with sampling at two locations ., To investigate this hypothesis , we compared coding schemes in which the same number of bits ( four bits at each of two spatial samples ) were allocated to one , two , or four samples in time , and in which the spatial sampling was across the flow axis ( as in Fig 3 ) , or along the flow axis ., Fig 5 shows that this hypothesis is supported ., Considering first bin boundaries based on histogram equalization , and sensor locations across the flow axis ( unshaded portions of plots in first row of Fig 5 ) , two or more bits were only beneficial for the most diffusive environment boundary flow ( Fig 5C ) ., Likewise , for sensor locations along the flow axis ( shaded half of each subplot ) , more than one bit of resolution was only helpful in this environment ( boundary flow ( Fig 5C ) ) ., Similar conclusions are reached when bin boundaries are determined via the “greedy” binning procedure: more than one bit of resolution for odor concentration is only useful in the most diffusive environment ( boundary flow ( Fig 5C ) ) , and has the greatest benefit when the two sensors are across to the axis of flow ., In the fast flow condition , increasing resolution while decreasing the number of samples in time makes little difference ( Fig 5A ) , and for the slow flow condition ( Fig 5B ) , increasing resolution while decreasing the number of samples leads to a loss of information about location for either sensor orientation ., In sum , the results of Figs 4C , 5C1 and 5C2 show that in a diffusive regime the exact choice of bin boundaries is not important , but devoting up to four bits to concentration resolution has a benefit over accumulating multiple temporal samples ., When the flow conditions are more turbulent , a navigator benefits from classifying multiple odor samples at coarser resolution ( Fig 5A and 5B ) , but the choice of the discretization threshold becomes important ( Fig 4A and 4B ) ., Consistent across conditions , sampling across the odor plume yielded more MI than sampling along the mean flow direction ( white vs . gray shaded regions in Fig 5 ) ., We now discuss the implications of our findings , first with regard to sensation and then with regard to navigation algorithms ., As a starting point , we consider the simple scenario of a sensory system confronted with a continuous and widely varying input , but limited in the number of symbols that it can use for encoding ., As is well-known , information is maximized when each of the symbols is used equally often , i . e . , histogram equalization ., Histogram equalization can be implemented as a nonlinearity applied to the input prior to producing a neural output 18 ., For a positively skewed distribution , such as light intensities or odor intensities , the nonlinearity is a highly compressive one , so that it takes into account the rarity of very large inputs ., Here , however , we consider the task of maximizing information not about the sensory signal itself , but about location—which is related to odor concentration in a complex , stochastic manner ., As we showed , most of the available information about location can be conveyed by a coarse discretization of the sensory range—in fact , by binarization ., However , this only holds if the cutpoint is properly chosen ., In the two more turbulent odor environments considered here , the optimal cutpoint is substantially higher than the median , which is the cutpoint associated with histogram equalization ( see Fig 4 ) ., That is , discriminations in the upper range of odor concentrations play a disproportionately greater role in determining location , than in reconstructing the input per se ., Correspondingly , implementation of this encoding requires a nonlinearity that is less compressive for higher intensities than histogram equalization ., Optimal adaptation strategies , in the sense of being maximally informative , under naturalistic stimuli are ( to our knowledge ) unknown ., The problem of optimally discretizing a signal is not just an olfactory problem but applies to other sensory modalities which face resource constraints as well ( e . g . vision 26–28 ) ., While it is difficult to imagine a biologically-plausible mechanism that achieves the precisely optimal nonlinearity for conveying information about location , there is a simple and plausible mechanism that can achieve an approximation: ligand-receptor binding in olfactory receptor neurons 29 ., In steady-state , this mechanism generates a nonlinear encoding described by the Hill equation 30 ., This transformation compresses signals at high concentrations , because receptors become occupied , and more ligand is required to activate the remaining receptors 31 ., Thus , the degree of compression depends on the apparent dissocation constant Kd , the odorant concentration at which half of the receptors are occupied ., Setting Kd at the median odor concentration corresponds to histogram equalization: half of the time the ligand binding will be below the median , and half of the time it will be above ., Interestingly , setting Kd at the mean concentration , rather than the median , leads to less compression than histogram equalization ., This is because the measured odor concentrations are positively skewed ., Since the mean odor concentration is larger than the median , this setting will produce a response that is less than half-maximal most of the time ., Such a coding strategy results in more information about location than histogram equalization , as we have outlined above ( see Fig 4 ) ., In order to implement this strategy , olfactory receptors or receptor neurons would have to have an apparent Kd close to the mean concentration in the environment ., Adaptation of Kd to the mean has been observed in olfactory receptor neurons of the fruitfly 32–34 , and might serve to increase the amount of information that the fly olfactory system can encode about its location in a turbulent environment ., With regard to odor navigation algorithms , we note that these fall into two categories: those that rely on local cues ( e . g . comparison of concentration differences in two sensors 35 , comparison of sample arrival times in two sensors 13 , the combination of local anemotactic and olfactory cues 36 , 37 ) , and those algorithms that construct a cognitive map ( like infotaxis 1 and mapless 2 ) ., We do not intend to argue for one kind of strategy over the other , but rather to identify aspects of the odor navigation problem that apply to both , as both begin with the acquisition of sensory samples ., Our work suggests that these algorithms can operate on a coarse representation of odor concentration since we find that a four-bit representation of the odor intensity reveals almost the same amount of information as finer odor concentration representation ., We also found that sampling with two sensors adds substantially to the amount of information about location , and this improvement is not just due to obtaining two samples , but by comparing them in a labelled fashion ( as observed in the second row of Fig 3 ) ., While this is directly exploited by comparison algorithms using two sensors , we suggest that , navigation algorithms that use an internal model of the odor distribution like infotaxis and mapless could also be improved by incorporating measurements from two sensors ., Finally , an important caveat of our study is that animals have multi-sensory cues available; here we only consider the single modality of odor and do not integrate information of other modalities , e . g visual or mechanosensory flow information , that navigators have access to ., In particular , it is crucial for moths and fruitflies to combine flow information via mechanosensory input when walking and visual input when flying for successful navigation 38–41 ., For example , since the wind direction may meander substantially , a simple upwind movement can lead a navigator out of the odor plume 8 , 42 ., Simultaneously recording flow and odor concentration , and analysis along the lines undertaken here , may shed light on useful sampling strategies for combining both sources of information ., Determining the location of an odor source based on olfactory cues is a challenging problem ., We focused on how to optimally sample from the odor distribution when the goal is to determine location with respect to the source ., This study shows that the sampling strategy that maximizes information about location under finite resources utilizes two sensors , allowing for the comparison of spatially separated samples , while representing odor concentration in no more than three to four bits ., Furthermore , temporal sequences of samples can be averaged to preserve resources while only minimally affecting the amount of information that the sequence conveys . | Introduction, Methods, Results, Discussion | Many species rely on olfaction to navigate towards food sources or mates ., Olfactory navigation is a challenging task since odor environments are typically turbulent ., While time-averaged odor concentration varies smoothly with the distance to the source , instaneous concentrations are intermittent and obtaining stable averages takes longer than the typical intervals between animals’ navigation decisions ., How to effectively sample from the odor distribution to determine sampling location is the focus in this article ., To investigate which sampling strategies are most informative about the location of an odor source , we recorded three naturalistic stimuli with planar lased-induced fluorescence and used an information-theoretic approach to quantify the information that different sampling strategies provide about sampling location ., Specifically , we compared multiple sampling strategies based on a fixed number of coding bits for encoding the olfactory stimulus ., When the coding bits were all allocated to representing odor concentration at a single sensor , information rapidly saturated ., Using the same number of coding bits in two sensors provides more information , as does coding multiple samples at different times ., When accumulating multiple samples at a fixed location , the temporal sequence does not yield a large amount of information and can be averaged with minimal loss ., Furthermore , we show that histogram-equalization is not the most efficient way to use coding bits when using the olfactory sample to determine location . | Navigating towards a food source or mating partner based on an animals’ sense of smell is a difficult task due to the complex spatiotemporal distribution of odor molecules ., The most basic aspect of this task is the acquisition of samples from the environment ., It is clear that odor concentration does not vary smoothly across space in many natural foraging environments ., Using data from three different naturalistic environments , we compare different sampling strategies and assess their efficacy in determining the sources’ location ., Our findings show that coarsely encoding the concentration of samples at separate sensors and/or multiple times provides more information than encoding fewer samples with higher resolution ., Furthermore , coding resources should be focused on discriminating rare high-concentration odor samples , which are very informative about the sampling location ., Such a nonlinear transformation can be implemented biologically by the receptor binding kinetics that bind odorants as a first stage of the sampling process ., A further implication is that animals as well as computational models of algorithms can operate efficiently with a coarse representation of the odor concentration . | classical mechanics, chemical compounds, fluid mechanics, quantum tunneling, neuroscience, organic compounds, turbulence, probability distribution, animal behavior, mathematics, odorants, materials science, zoology, quantum mechanics, thermodynamics, entropy, animal cells, behavior, materials by attribute, chemistry, fluid dynamics, olfactory receptor neurons, probability theory, continuum mechanics, physics, animal migration, cellular neuroscience, acetones, organic chemistry, cell biology, animal navigation, neurons, biology and life sciences, cellular types, physical sciences, afferent neurons | null |
journal.pgen.1004404 | 2,014 | Regulation of Gene Expression in Autoimmune Disease Loci and the Genetic Basis of Proliferation in CD4+ Effector Memory T Cells | Memory T cells are an important component of the adaptive immune system ., They circulate between lymphoid organs , blood , and peripheral tissues , and facilitate faster and more aggressive immune response to antigens after re-exposure ., CD4-positive effector memory T ( CD4+ TEM ) cells are known to migrate to peripheral sites of inflammation upon activation , and rapidly produce both Th1 and Th2 cytokines 1 ., Investigators have long suggested their involvement in autoimmune diseases including rheumatoid arthritis ( RA ) , type I diabetes ( T1D ) , and celiac disease ( CeD ) 2–5 ., However , whether changes in cell population subsets and functions are causal or reactive to disease is uncertain ., One strategy to answer this question is to examine potential intermediate molecular phenotypes , and identify those modulated by genetic variants ., In order to understand the pathogenic roles of CD4+ TEM cells in autoimmunity , we aimed to characterize the variation in their phenotypic and functional markers in a healthy population , and to identify whether these markers intersect with the genetic basis for autoimmunity ., The majority of autoimmune disease risk variants are located in non-coding regions of the genome ., It is reasonable to hypothesize that a subset of them causes disease by altering gene regulatory mechanisms as expression quantitative trait loci ( eQTL ) 6–9 ., So far , studies of gene regulation have largely been carried out in cell lines and primary resting blood cells including undifferentiated CD4+ T cells , B cells , monocytes , and dendritic cells 8 , 10–12 ., However , to understand the pathogenic mechanisms of risk variants , especially when studying the immune system where cells are highly diverse and functionally specialized , it is crucial to focus on relevant cell types and stimulated cellular states ., We have previously shown that genes within RA risk loci were most specifically expressed in CD4+ TEM cells , compared to more than 200 other immune cell types of various lineages and developmental stages ( p\u200a=\u200a1 . 00×10−8; Figure S1 ) 13 ., Celiac disease and T1D loci were also enriched for genes specifically expressed in CD4+ TEM cells ( p\u200a=\u200a1 . 43×10−5 and 1 . 29×10−4 , respectively; Figure S1 ) 13 ., Non-coding single nucleotide polymorphisms ( SNPs ) associated with RA significantly overlap chromatin marks of trimethylation of histone H3 at lysine 4 ( H3K4me3 ) specifically in CD4+ regulatory and memory T cells ( p\u200a=\u200a1 . 3×10−4 and 7 . 0×10−4 , respectively ) 14 ., We hypothesized that the risk alleles of these conditions might influence CD4+ TEM quantitative molecular phenotypes:, 1 ) the expression of immune-related genes;, 2 ) the relative abundance of CD4+ TEM cells in peripheral blood; and, 3 ) proliferative response to T cell receptor ( TCR ) stimulation ., To this end , we undertook a large immunoprofiling study in a healthy population of 174 European-descent individuals , by cross-analyzing genotype , transcription , abundance , and proliferative response in primary CD4+ TEM cells ., Because the post-stimulation activation of CD4+ TEM cells is presumably crucial for their autoimmune response , we assayed cells not only at rest , but also after T cell receptor ( TCR ) stimulation with anti-CD3/CD28 beads ., As such , this study is the first to our knowledge to map expression quantitative trait loci and examine immunological cellular traits in primary CD4+ TEM cells under multiple states ., Using the ImmunoChip platform , investigators recently densely genotyped 186 loci disease that originally arose through genome-wide association studies ( GWAS ) in case-control samples for RA , CeD , and inflammatory bowel disease 15–17 , as well as T1D ( unpublished data ) ., Dense genotyping allowed localization of association signals within these disease loci to a set of alleles that are very likely to be causal ., Within these loci , we have a greater ability to identify co-localization between alleles driving variation in molecular phenotypes ( such as eQTLs ) and the disease risk alleles ., However , in instances where multiple variants are in perfect linkage , we cannot pinpoint the exact causal variant without functional evaluation ., We first aimed to identify SNP variants that regulated expression of genes in cis ., To best localize eQTL signals , we imputed 1000 Genomes variants within 250 kb from the transcription start site ( TSS ) of each gene ( excluding five HLA genes and five long non-coding RNAs ) ., We tested SNPs in gene-coding and non-coding regions in both resting and stimulated CD4+ TEM cells ., We included gender and the top five principal components of the genotype data ( calculated by EIGENSTRAT ) as covariates in regression ., To adjust for multiple hypothesis testing , we conducted 10 , 000 permutations within each gene region to calculate empirical p-values , and then reported associations at a false discovery rate of 5% ., In total , we observed 46 genes ( 22 . 4% ) with cis-eQTL signals , including 17 in resting cells and 43 in stimulated cells ( Tables 1 and 2 , Figure 2A ) ., For 14 of the 46 genes ( 30 . 4% ) , we detected eQTL signals in both resting ( 14/17 , 82 . 4% ) and stimulated ( 14/43 , 32 . 6% ) states ., In four of these 14 genes ( FHL3 , GRB10 , IL18R1 , and PIGC ) , the lead eQTL SNPs across resting and stimulated states were identical ., In another five genes ( C1QTNF6 , PRDM1 , SKAP2 , DDX6 , and LYRM7 ) , the lead SNPs are in tight LD ( r2\u200a=\u200a0 . 80∼1; based on 1000 Genomes Release 2 , European samples ) ., For the remaining five genes ( BLK , TMPRSS3 , CD101 , ORMDL3 , and GSDMB ) , the lead SNPs from the two states were in partial LD ( 0 . 42<r2<0 . 56 ) ., In these five cases , we could not be confident that the eQTL SNPs across stimulation states were tagging the same variant ., Three genes ( IL23R , PLCH2 , and RGS1 ) had statistically significant eQTLs exclusively in the resting state , while 29 genes had statistically significant eQTLs exclusively in stimulated cells , such as rs942793 associated with ZMIZ1 expression ( Figure 2B ) ., One possibility is that some SNPs failed to reach significance threshold due to the small sample size or low expression levels in resting cells ., However , we observed many genes with truly state-specific eQTLs , where the estimated effect sizes ( β ) of the eQTL SNP differed significantly across resting and stimulated states ., To systematically compare the βrest and βstim for each gene , we used a z-statistic to quantify the probability that they differ ., We then reported the p-value ( two-tailed ) assuming that z is distributed as standard normal , considering p<0 . 05 to be significantly different ( “state-specific”; see Tables 1 and 2 ) ., For example , rs12746918T increased the expression of PLCH2 significantly only in resting cells; and βrest was approximately twice as large as βstim ( Figure 2C ) ., We note that 1 of the 3 eQTLs in resting cells was state-specific ( p<0 . 05 ) , and 13 out the 29 eQTLs seen in stimulated cells were state-specific ( p<0 . 05 ) ., Of the 14 eQTLs that were shared between resting and stimulated cells , only 4 of them , BLK ( Figure 2D ) , CD101 , PIGC , and PRDM1 , had different βs across states ., The abundance of eQTLs detected exclusively in stimulated cells underscores the importance of studying cells in different cellular states ., We wanted to assess whether the eQTLs might act by altering gene regulatory elements in CD4+ TEM cells ., To this end we asked whether the eQTL SNPs co-localized with marks of active promoters or enhancers ., We utilized H3K4me3 marks from the NIH Roadmap Epigenomics Mapping Consortium 20 measured by ChIP-seq in primary CD4+ memory T cells ., For the SNP with the strongest association to each gene , we queried the distance of the nearest H3K4me3 mark to this SNP or its LD partners ( r2>0 . 8 ) ., We compared this distance measure between two sets of SNPs: the 46 SNPs with significant eQTL associations ( FDR<5% , resting or stimulated ) , and the SNPs most strongly correlated with the other 159 genes but did not reach significance threshold ., Indeed , the 46 significant eQTL SNPs were located at smaller distances to H3K4me3 marks ( p\u200a=\u200a1 . 10×10−7 , one-sided Mann-Whitney test , Figure S3A ) ., In addition , we queried the height of each H3K4me3 marks peak , which reflected the number of reads at a given position compared to genomic controls as defined by the MACS software package ., A tall peak gives us confidence that the mark is present in a large proportion of cells ., Comparing the marks nearest to the two sets of SNPs , we saw that the 46 eQTL SNPs were also located near taller peaks ( p\u200a=\u200a9 . 56×10−8 , Figure S3B ) ., We compared the cis-eQTLs we discovered to those found in heterogeneous peripheral blood mononuclear cells ( PBMC ) in a large genome-wide eQTL meta-study ( n\u200a=\u200a5 , 331 ) conducted by Westra et al . 8 ., At 5% FDR , eleven of the 46 eQTL genes we identified showed no detectable signal in PBMCs at 50% FDR ., We saw significant associations in 131 genes at 50% FDR , 53 of which had no signal in PBMCs at 50% FDR ( Tables 1 and 2 ) ., We hypothesized that these genes tended to be more specifically expressed in CD4+ TEM cells , thus making eQTLs readily detectable in the purified cell population ., To assess this , we examined cell-specific expression of the genes the ImmGen dataset , which assayed the genome-wide expression in 247 murine mouse immunological cell types 13 , 21 ., We found that the genes with CD4+ TEM cell-specific eQTLs ( at 50% FDR ) were more specifically expressed in CD4+ TEM cells than genes with eQTLs detected in both datasets ( p\u200a=\u200a0 . 044 , one-sided Mann-Whitney test ) ., We then focused on 115 genes near 96 risk alleles of RA , T1D , and/or CeD in densely genotyped loci ( 182 gene-SNP pairs , including two risk alleles shared by at least two diseases , see Tables S2 and S3 ) ., We discovered that eleven ( 11 . 4% ) disease-associated SNPs ( 6 of 24 RA SNPs , 5 of 37 T1D SNPs , and 3 of 37 CeD SNPs ) correlated significantly with the expression of ten genes in either resting or stimulated state ( Table S3 ) ., In addition , there was substantial enrichment of nominally significant associations ( p<0 . 05 ) among disease SNPs ., By random chance , we expected about nine SNP-gene pairs to reach nominal association in each stimulation state ., However , we observed 26 pairs ( 14 . 2% ) with nominal association in resting cells ( p\u200a=\u200a4 . 67×10−7 , one-tailed binomial test ) ., Even more strikingly , we observed 45 pairs ( 24 . 7% ) with nominal association in stimulated cells ( p<10−15 , one-tailed binomial test ) ., To identify those instances where the disease-associated SNP could explain the entire eQTL signal in the gene region , we applied conditional analysis to identify any residual signals after controlling for the disease SNP ., In five of the ten genes ( BLK , C5orf30 , GSDMB , IRF5 , PLEK ) , conditioning on the disease SNP obviated any remaining eQTL signal in the region ( no SNP with permutation p-value <0 . 05; Figure 3 ) , suggesting that there was a single variant ( the disease-associated SNP or one in very high LD to it ) that drove variation in expression ., Interestingly , as previously noted , the lead SNPs in resting and stimulated states for BLK and GSDMB were in partial linkage to each other ., The absence of residual eQTL signal upon conditioning on the same risk allele might suggest that the lead SNPs were indeed tagging the same causal SNP in each of these genes ., In each of the other five genes ( ORMDL3 , SKAP2 , TMPRSS3 , TNFRSF14 , and ZMIZ1 ) , evidence of independent eQTL effect remained after conditional analysis ., In these instances the disease-associated SNP and remaining lead signal are in partial linkage disequilibrium ( r2\u200a=\u200a0 . 36–0 . 73 ) ., In these cases , we could not conclude whether the disease SNPs drove the alteration in expression , or whether the true causal SNPs were in partial linkage and caused spurious associations ., It is probable that disease risk alleles were indeed causal , yet we could not confidently fine-map the effect due to experimental noise in expression assays or inadequate sampling ., We note that another 26 genes within disease loci associated contained cis-eQTL signals , but that these cis-eQTL signals did not co-localize with RA , T1D , or CeD alleles ., As these loci had been fine-mapped using Immunochip , the lack of overlap strongly suggested that these cis-eQTLs and disease-causing variants were distinct ., For example , rs798000 is an RA risk allele located in a non-coding region upstream of CD2 , CD58 , and PTGFRN ., However , it was not associated with the expression of any of these genes ( p>0 . 5 ) ., Another example was rs6911690 , an RA allele located about 60 kb 5′ of PRDM1 , that was not associated with the expression of the gene at rest or after stimulation ( p>0 . 5 ) ., The lead eQTL SNP associated to PRDM1 was rs578653 ( FDR<10−3 ) , which was not in LD with the disease allele ( r2<0 . 05 ) ., The relative peripheral abundance of CD4+ TEM cells varied between individuals ( mean\u200a=\u200a9 . 57%; SD\u200a=\u200a4 . 85% ) , and was reproducible 35 individuals with two separate blood draws more than one month apart ( Pearsons r\u200a=\u200a0 . 87 , p\u200a=\u200a1 . 77×10−11 , see also Figure S2B ) ., Consistent with other studies , we observed that the relative proportion of CD4+ TEM cells increased with age by 0 . 11% per year ( page\u200a=\u200a1 . 92×10−3 ) 22 ., We also observed that on average men had 2 . 22% more CD4+ TEM cells than women ( pgender\u200a=\u200a3 . 80×10−2; see Figure S4 ) ., Upon anti-CD3/CD28 stimulation , there was a substantial inter-individual variation in proliferation measured by both division index ( DI , average number of divisions undergone by all cells; mean\u200a=\u200a1 . 46 , SD\u200a=\u200a0 . 35 ) , and proliferation index ( PI , average number of divisions undergone only by dividing cells; mean\u200a=\u200a2 . 16 , SD\u200a=\u200a0 . 21 ) ., Proliferation metrics were also reproducible in the 35 individuals ( Pearsons rDI\u200a=\u200a0 . 57; Pearsons rPI\u200a=\u200a0 . 62 , Figures S2C and S2D ) ., Interestingly , proliferation was negatively correlated to the proportion of CD4+ TEM cells ( pDI\u200a=\u200a1 . 28×10−3 , pPI\u200a=\u200a1 . 93×10−3 ) , but was not associated to age or gender ( p>0 . 3 ) ., This negative correlation needs to be replicated in an independent dataset ., Effector functions of TEM cells with higher proliferative capacities need to be examined to understand whether they represent a hyperactive subset whose abundance is controlled to maintain immune homeostasis ., Possibly individuals with a lower proportion of TEM cells are relatively enriched for these subsets ., We tested genome-wide SNPs for association to relative abundance , division index , and proliferation index , considering p<5×10−8 as the threshold for significance ., For abundance , we included gender , age , and the top five principal components of genotypes as covariates ., Given the correlation with proliferation , we also included the measured CD4+ TEM relative abundance as an additional covariate ., We observed associations to division index in several loci , including 13q34 led by rs389862 ( p\u200a=\u200a4 . 75×10−8; Figure 4A ) ., This SNP is a non-coding variant located 30 kb upstream of RASA3 , and 70 kb upstream from CDC16 ., Both genes have known roles in regulating cell proliferation or differentiation 23 , 24 ., This SNP was also strongly associated with proliferation index ( p\u200a=\u200a2 . 75×10−7 ) ., Additionally , there was a strongly suggestive association to rs3775500 on chromosome 4 , located in the intron of DAPP1 , which encodes the Bam32 protein ( p\u200a=\u200a5 . 40×10−7; Figure 4B ) , which is an adaptor protein expressed solely in antigen presenting B cells ., Interestingly , mutations in this gene have been shown by several groups to affect T cell activation 25 , 26 , suggesting the possibility that B cells may indirectly regulate T cell function in autoimmunity ., We did not observe any significant association with the relative abundance of CD4+ TEM cells ., When we extracted the association statistics of 118 densely genotyped risk alleles of CeD , RA , and/or T1D , they showed no inflation in association p-values for relative abundance of CD4+ TEM cells ( Figure 5A , Table S2 ) ., This suggested that risk variants did not modify risk via modulation of CD4+ TEM peripheral abundance ., We recognized that the power to detect significant associations might have been limited in our study by the sample size ., However , this negative finding was corroborated by results from a recently published study with data from ∼2800 individuals , in which the same set of risk alleles also showed no significant association to CD4+ TEM ( see Figure S5 ) 27 ., Similarly , the same set of risk alleles did not show significant association to proliferative response ( Figure 5B , Table S2 ) ., Based on these data , it was unlikely that SNP variants associated to RA , T1D , or CeD conferred risk through modulation of CD4+ TEM cell abundance or proliferation ., After stimulation we observed that 122 genes showed significant changes in expression in response to stimulation , including 78 whose expression at least doubled or decreased by 50% ( Table S1 ) ., The gene with the greatest post-stimulation induction was GZMB ( average fold change\u200a=\u200a93 . 48 ) , which encodes granzyme B , a protein involved in the apoptosis of target cells during cell-mediated immune response in cytotoxic and memory lymphocytes ., The most significantly down-regulated gene was GRB10 ( average fold change\u200a=\u200a0 . 18 ) , which is near rs6944602 associated with T1D and encodes growth factor receptor-bound protein 10 , whose function in the immune system is unclear ., We observed that relative gene expression at rest predicted proliferative response ., In 182 individuals with both proliferation and gene expression data , 17 of the 215 genes were associated with proliferation index ( p<0 . 01 , two-tailed test by permuting proliferation data , Figure 6 , Table S1 ) ., Increased expression of 15 of the 17 genes including CCR5 , IL2RB , PRR5L , and TBX21 , were correlated with reduced proliferative response , while CCR9 and the lncRNA XLOC_003479 showed significant correlation with increased proliferation ., This number of correlated genes was far in excess of random chance based on a null distribution consisting of 1000 permutations ( p<10−3 , median 2 , maximum 15 ) ., The weighted sum of the 17 genes served as a “proliferation potential signature” , where we weighted the positively- and negatively-correlated genes as +1 and −1 , respectively ., This signature strongly predicted proliferation index ( r\u200a=\u200a0 . 55 ) ., We show the correlation between each of the 17 genes as well as the aggregate signature to proliferation as a heatmap ( Figure 6A ) ., To assess if we were overfitting the data , we applied a two way cross-validation , where we defined the proliferation signature based on genes from half of the individuals and tested correlation to proliferation in the remaining half of the individuals ., In both instances we again observed significant prediction of proliferation ( r\u200a=\u200a0 . 41 , one tailed p<10−3 by permutation; r\u200a=\u200a0 . 39 , p<10−3 ) ., To search for biological pathways underlying genes correlated to proliferation , we applied gene set enrichment analysis ( GSEA ) to test for enrichment for 1 , 008 functional gene sets based on Gene Ontology codes 28 ( Figure 6B ) ., Genes correlated to reduced proliferation were most significantly enriched for GO:0012502 ( induction of programmed cell death; one tailed p\u200a=\u200a1 . 8×10−4 ) ; those correlated with increased proliferation were most significantly enriched for GO:0002285 ( lymphocyte activation involved in immune response , one tailed p\u200a=\u200a3 . 9×10−4 ) ., Using data from 29 individuals each with two samples collected at least one month apart , we replicated the observed correlation ., In these samples we performed a cross-visit analysis , and observed that the same 17-gene signature from the first visit significantly predicted proliferation indices on the second visit ( r\u200a=\u200a0 . 65 , p\u200a=\u200a0 . 0006 , 1-tailed permutation ) , and vice versa ( r\u200a=\u200a0 . 55 , p\u200a=\u200a0 . 0019 ) ., To fine-map and link risk loci to their pathogenic mechanisms , we investigated molecular and immune phenotypes potentially leading to disease end-points ., The immune system is particularly complex , and different cells under various activation states have specialized functions that may not be adequately captured by examining PBMCs ., Therefore , we focused on one purified cell population that had been shown to be important for the pathogenesis of several autoimmune diseases ., We quantified population variation in several traits , including peripheral abundance , proliferative response to TCR stimulation , and expression of genes within autoimmune disease loci at rest and after stimulation ., In Tables 1 , 2 , and S2 , we provide significant cis-eQTLs and genome-wide association results ., To our knowledge this study was a first cross-examination of genetic- , transcriptional- , and cellular-level quantitative traits in CD4+ TEM cells ., It demonstrated the importance of focusing functional studies in a purified cell population under relevant developmental and stimulation states ., By examining the proliferative response upon TCR stimulation , we identified a subset of genes whose baseline expression predicted proliferative potential ., Intriguingly , these genes were involved in programmed cell death and lymphocyte activation ., Whether variation in proliferative abilities correlated with cytokine production and other signaling functions , thus affecting susceptibility to autoimmunity , remains a question to be addressed by future studies ., Of the 205 genes in disease loci that we examined , 46 had cis-eQTLs ., Notably , eleven of these were specific to stimulated CD4+ TEM cells , and not previously found in PBMCs ., We noted that approximately 10% of genes within risk loci of diseases had cis-eQTLs ., However in many instances the lead eQTL SNPs were unrelated to the disease-associated SNPs ., One example of a disease allele that functioned as cis-eQTL was rs39984 , which was associated to lower risk of RA , and regulated the expression of C5orf30 encoding an UNC119-binding protein ., This SNP variant is located in the first intron of C5orf30 , and indeed explained the entire cis-eQTL signal in this gene ( see Figure 2B ) ., This eQTL effect was previously undetected in PBMCs , and the proteins functional role in the immune system is largely unknown ., However , a recent study showed that rs26232 ( the lead GWAS SNP prior to fine-mapping , r2\u200a=\u200a0 . 988 to rs39984 ) was correlated with lower severity of radiologic damage in RA , independent of previously established biomarkers 29 ., Another gene in the locus , GIN1 , is located 140 kb from rs39984; however its expression showed no correlation with the SNP ( p>0 . 5 ) ., Another CD4+ TEM cell-specific eQTL gene was DDX6 , which encodes DEAD-box RNA helicase 6 ., However , in this case , the lead eQTL SNP ( rs4938544 ) associated to increased expression of DDX6 in stimulated cells was not in LD with the known CeD risk allele ( rs10892258 , r2<0 . 1 ) or the RA risk allele ( rs4938573 , r2<0 . 1 ) ., Neither risk allele showed significant association to DDX6 expression ( p\u200a=\u200a0 . 19 and 0 . 26 , respectively ) ., Both risk alleles are also located near CXCR5 , BCL9L , and TREH; none of these genes had reported cis-eQTLs in PBMCs 8 ., However , we did not assay these three genes in this study , therefore could not confirm the role of disease alleles in regulating their expression in CD4+ TEM cells ., Although we did not assay all genes or test for trans-acting eQTLs , based on the level of co-localization between eQTL SNPs and risk alleles observed in the study , we found it unlikely that all non-coding risk variants caused disease by altering gene expression within resting or stimulated CD4+ TEM cells ., In addition , while changes in proportions of lymphocyte subsets had been observed in patients of autoimmune disorders 30–35 , we did not find evidence to support disease alleles roles in directly modulating CD4+ TEM cell abundance or proliferative response ., Ultimately , other cell states and cell types will need to be investigated ., We recognize several limitations to the current study ., In order to conduct a focused study on a small amount of purified primary cells we used the NanoString nCounter assay system ., This avoided potential biases and artifacts arising from cDNA synthesis required for microarray or RNA-seq studies , but restricted our analysis to a subset of candidate genes within risk loci of CeD , RA , and T1D , rather than a genome-wide expression analysis ., Consequently we could not identify trans-eQTLs , splice variants , or epistatic effects on expression regulation ., Additionally , anti-CD3/CD28 stimulation for memory T cells is not antigenic , especially while in isolation from a “natural” multi-cellular environment , thus it was only partially physiological ., This and other cell-specific studies on population variation in molecular phenotypes are only a beginning of examining potential intermediate phenotypes ., Post-activation cytokine production by CD4+ TEM cells are likely crucial in driving autoimmunity ., Therefore , it is critical that future studies of molecular phenotypes include proteomic assays to quantify functional markers of immune response ., Finally , functional experiments will need to be conducted in the future to determine whether these molecular phenotypes are indeed intermediary to disease ., All research was approved by our Institutional Review Board , and informed consent was obtained from each volunteer ., We enrolled 225 healthy volunteers ( 134 females , 91 males ) of non-Hispanic Caucasian descent that proved informed consent through the Phenogenetics Project at Brigham and Womens Hospital ., Subjects ages ranged from 19 to 57 years with average female and male ages of 28 . 8 years and 34 . 9 years , respectively ., Thirty-five subjects ( 18 females , 17 males ) returned for a second study visit one to nine months after their initial visits ., We genotyped each subject using the Illumina Infinium Human OmniExpress Exome BeadChip ., In total , we genotyped 951 , 117 SNPs , of which 704 , 808 SNPs are common variants ( minor allele frequency MAF>0 . 01 ) and 246 , 229 are part of the exome ., After quality control , 638 , 347 common SNPs remained ., Of all subjects , 174 subjects had abundance , proliferation , gene expression , and quality controlled genotype data ., Detailed quality control criteria are described in Text S1 ., For each gene , we selected a 500 kb region ( 250 kb each in the 3′ and 5′ directions ) around the transcription start site and imputed 1000 Genomes SNPs into the genome-wide SNP data using BEAGLE Version 3 . 3 . 2 ., We used the European samples from 1 , 000 Genomes as the reference panel ., We excluded markers that had MAF<0 . 05 in the reference panel as well as all insertion/deletions ., After imputation , we excluded markers with a BEAGLE R2<0 . 4 or MAF<0 . 01 in the imputed samples ., We isolated peripheral blood mononuclear cells ( PBMC ) from whole blood using a Ficoll density gradient ( GE Healthcare ) ., We then isolated CD4+ effector memory T cells from PBMCs first by magnetic-activated cell sorting to enrich for CD4+ T cells , followed by fluorescent-activated cell sorting using labeled antibodies against CD45RA , CD45RO , and CD62L ., We stimulated CD4+ TEM cells by incubation with commercial anti-CD3/CD28 beads for 72 hours ., For proliferation studies , we labeled cells with carboxyfluorescein diacetate succinimidyl ester ( CFSE; eBioscience ) , and measured proliferation by dye dilution ., Detailed isolation and purification methods are described in Text S1 ( also see Figure S2A ) ., We designed the NanoString codeset based on GWAS SNPs associated with CeD , RA , and T1D as of April 2012 ., This list of SNPs can be found in Supplementary Table S4 ., As the numbers of associated loci with autoimmune diseases continuously expand , we refer the reader to ImmunoBase ( https://www . immunobase . org ) for up-to-date disease regions ., For each locus , we defined a region of interest implicated by the GWAS lead SNP 36 ., We identified the furthest SNPs in LD in the 3′ and 5′ directions ( r2>0 . 5 ) ., We then extended outward in each direction to the nearest recombination hotspot ., If no genes were found in this region , we extended an additional 250 kb in each direction ., All genes overlapping this region were considered implicated by the locus ., The final NanoString codeset ( prior to expression data quality control ) included 312 genes , including 270 genes near SNPs associated with 157 RA , CeD , and T1D through GWAS , 26 genes of immunological interest , and 15 reference genes with minimal change in expression after TCR stimulation ( see Supplementary Table S1 ) ., After quality control , 215 genes remained ., Of all 225 subjects in the study , 187 subjects passed gene expression quality control for both resting and stimulated cells ., Specific normalization and quality control procedures are described in Text S1 ., To control for any potential population stratification , we adjusted all association tests using the top five principal components of our genome-wide SNP data ., Principal components were generated via EIGENSTRAT using unsupervised analysis ( no reference populations were used ) ., The top five PCs explained 6 . 88% ( 2 . 08% , 1 . 27% , 1 . 20% , 1 . 17% , and 1 . 16% , respectively ) of the total variance ., After controlling for these five PCs , the lambda GC for CD4 TEM proportion association was 1 . 008; that of division index was 1 . 001 ., For each gene-SNP pair , we applied linear regression using the first five principal components of the genotype data and gender as covariates ., As such , normalized expression\u200a=\u200aβ0+β1*allelic dosage+β2*PC1+β3*PC2+β4*PC3+β5*PC4+β6*PC5+β7* ( factor ) gender ., To adjust for multiple hypothesis testing while taking into consideration the correlation among SNPs within each locus , we calculated a permutation-based p-value for each SNP ., We performed 10 , 000 permutations of the residual expression values ., We reported each SNPs p-value the proportion of permutation P value smaller than the analytical p-value ., For conditional analysis , the vector of allelic dosages of the disease-associated SNP was included as an additional covariate ., We defined CD4+ TEM cells as CD45RA− , CD45RO+ , and CD62Llow/− ., In all samples CD4+ TEM cells were quantified using X-Cyt , a mixture-modeling based clustering program for automated cell population identification ( see Figure S6 ) 19 ., We fit proliferation division peaks with one-dimensional Gaussian mixture models ( see Figure S7 ) ., Detailed protocol and algorithms are described in Text S1 ., All linkage disequilibrium calculations ( r2 ) were based on 1000 Genomes Release 3 European samples ., All association tests were performed using Plink v1 . 07 ., We considered p<5×10−8 to be genome-wide significant; p<5×10−5 was considered as suggestive ., CD4+ TEM abundance and proliferation correlations with age and gender were calculated by multivariate linear model implemented in R-3 . 0 ., We calculated two-sample comparisons ( CD4+ TEM cell-specific expression between genes , and H3K4me3 h/d scores between SNPs ) with the Mann-Whitney test ., Details of statistical analyses are described in Text S1 ., We make all phenotypic data ( expression , periphera | Introduction, Results, Discussion, Materials and Methods | Genome-wide association studies ( GWAS ) and subsequent dense-genotyping of associated loci identified over a hundred single-nucleotide polymorphism ( SNP ) variants associated with the risk of rheumatoid arthritis ( RA ) , type 1 diabetes ( T1D ) , and celiac disease ( CeD ) ., Immunological and genetic studies suggest a role for CD4-positive effector memory T ( CD+ TEM ) cells in the pathogenesis of these diseases ., To elucidate mechanisms of autoimmune disease alleles , we investigated molecular phenotypes in CD4+ effector memory T cells potentially affected by these variants ., In a cohort of genotyped healthy individuals , we isolated high purity CD4+ TEM cells from peripheral blood , then assayed relative abundance , proliferation upon T cell receptor ( TCR ) stimulation , and the transcription of 215 genes within disease loci before and after stimulation ., We identified 46 genes regulated by cis-acting expression quantitative trait loci ( eQTL ) , the majority of which we detected in stimulated cells ., Eleven of the 46 genes with eQTLs were previously undetected in peripheral blood mononuclear cells ., Of 96 risk alleles of RA , T1D , and/or CeD in densely genotyped loci , eleven overlapped cis-eQTLs , of which five alleles completely explained the respective signals ., A non-coding variant , rs389862A , increased proliferative response ( p\u200a=\u200a4 . 75×10−8 ) ., In addition , baseline expression of seventeen genes in resting cells reliably predicted proliferative response after TCR stimulation ., Strikingly , however , there was no evidence that risk alleles modulated CD4+ TEM abundance or proliferation ., Our study underscores the power of examining molecular phenotypes in relevant cells and conditions for understanding pathogenic mechanisms of disease variants . | Genome-wide association studies have identified hundreds of genetic variants associated to autoimmune diseases ., To understand the mechanisms and pathways affected by these variants , follow-up studies of molecular phenotypes and functions are required ., Given the diversity of cell types and specialization of functions within the immune system , it is crucial that such studies focus on specific and relevant cell types ., Here , we studied genetic and cellular traits of CD4-positive effector memory T ( CD4+ TEM ) cells , which are particularly important in the onset of rheumatoid arthritis , celiac disease , and type 1 diabetes ., In a cohort of healthy individuals , we purified CD4+ TEM cells , assayed genome-wide single nucleotide polymorphisms ( SNPs ) , abundance of CD4+ TEM cells in blood , proliferation upon T cell receptor stimulation , and 215 gene transcripts in resting and stimulated states ., We found that expression levels of 46 genes were regulated by nearby SNPs , including disease-associated SNPs ., Many of these expression quantitative trait loci were not previously seen in studies of more heterogeneous peripheral blood cells ., We demonstrated that relative abundance and proliferative response of CD4+ TEM cells varied in the population , however disease alleles are unlikely to confer risk by modulating these traits in this cell type . | blood cells, genome-wide association studies, genome expression analysis, rheumatology, medicine and health sciences, functional genomics, immune cells, immunology, genome analysis, epigenomics, autoimmunity, white blood cells, animal cells, t cells, immune response, genetics of the immune system, cell biology, clinical immunology, transcriptome analysis, genetics, biology and life sciences, cellular types, genomics, genetics of disease, computational biology | null |
journal.pcbi.1000093 | 2,008 | Machine-Learning Approaches for Classifying Haplogroup from Y Chromosome STR Data | Genetic variation on the non-recombining portion of the Y chromosome ( NRY ) has become the target of many recent studies with applications in a variety of disciplines , including DNA forensics 1 , 2 , medical genetics 3 , genealogical reconstruction 4 , molecular archeology 5 , non-human primate genetics 6 , and human evolutionary studies 7–9 ., Two extremely useful classes of marker on the NRY include microsatellites or short tandem repeats ( STRs ) and single nucleotide polymorphism ( SNPs ) 7 ., STRs consist of variable numbers of tandem repeat units ranging from 1 to 6-bp in length and mutate via a stepwise mutation mechanism , which favors very small ( usually one repeat unit ) changes in array length ., Because high mutation rates ( estimated to be 0 . 23%/STR/generation ) in human pedigrees 10 , 11 often lead to situations where two alleles with the same repeat number are not identical by descent , STRs are not the marker of choice for constructing trees or for inferring relationships among divergent human populations ., Rather , the high heterozygosity of STRs makes them useful for forensic and paternity analysis , and for inferring affinities among closely related populations ., Reconstructing relationships among globally dispersed populations or divergent male lineages requires polymorphisms with lower probabilities of back and parallel mutation ( i . e . , lower levels of homoplasy ) and systems for which the ancestral state can be determined ., SNPs and small indels , with mutation rates on the order of 2-4×10−8/site/generation , are best suited for these purposes ., Because SNPs and indels are likely to have only two allelic classes segregating in human populations , they are sometimes referred to as binary markers ( we refer to both classes of marker as SNPs ) ., The combination of allelic states at many SNPs on a given Y chromosome is known as a haplogroup ., A binary tree of NRY haplogroups with a standard nomenclature system ( Figure S1 ) has been published and widely accepted among workers in the Y chromosome field 12 , 13 ., This Y chromosome tree is characterized by a hierarchically arranged set of 18 arbitrarily defined clusters of lineages ( clades A–R ) , each with several sub-clades ., By typing informative sets of SNPs , it is possible to assign samples to particular clades or subclades 9 , 14 ., One of the challenges for geneticists is the cost and time typically needed to genotype an appropriate number of SNPs to assign a given Y chromosome to a haplogroup ., Multiplex strategies to type SNPs are also difficult and require a substantial initial investment to implement 15 ., STRs on the NRY ( Y-STRs ) offer an alternative method for inferring the haplogroup of a sample ., It has been recognized for some time that STR variability is partitioned to a greater extent by differences among haplogroups than by differences among populations 16 , 17 ., This suggests that Y-STRs contain information about the haplogroup status of a given Y chromosome ., Because many Y-STRs can be genotyped in multiplex assays , typing appropriate sets of Y-STRs could represent a cost effective strategy for classifying Y chromosomes into haplogroups ., In this paper , we assess this possibility from a computational perspective and show how a suite of modern machine learning algorithms can automatically classify and predict haplogroups based on allelic data from a suite of Y-STRs ., We adapt three types of classifiers based on both generative and discriminative models to this problem ., When all the methods agree in tandem , we combine the classifications from each into a haplogroup assignment ., This enables an automatic , high throughput analysis pipeline for determining the haplogroup of a large number of samples in a cost effective and accurate manner ., We evaluated the performance of each classifier individually and in tandem using cross-validation on our 8 , 414 sample ground-truth training set , and compared the results with the nearest neighbor heuristic previously mentioned ., We also performed cross-validation on publicly available data from other published research with Y-STR and haplogroup data ., Finally , we tested the classification performance on the public data using our data for training ., In brief , the results show the classifiers perform very well with a diverse training set and that the number of loci available in the data set is an important determining factor in their performance ., The cross-validation was accomplished by stochastically partitioning the data sets into k equally sized subsets , iteratively holding out each one while training on the remaining data , and then testing on the held out subset ., More formally , let the ground-truth data set with N samples be ., We create equally sized subsets for 1≤i≤k that form a partition of , i . e . , ( 1 ) ( 2 ) We held out each subset Ai of the partition and trained the classifiers on the set ., A classification test was then performed on the held out set ., In practice , the subset sizes may differ by one if N/k is not integral ., For our experiments we chose to use k\u200a=\u200a5 folds ., The cross-validation was repeated 10 iterations , each time generating a random , equal partition of the data ., The performance results were finally compiled with the mean and standard error statistics ., We combined the classification output for a sample from the decision trees ( J48 and PART ) , Bayesian models , and support vector machines into a tandem decision ., The output haplogroups from each of the classifiers were compared together , and if they were in agreement , accepted , or assigned , the classification; otherwise the sample was left unassigned if they disagree and held-out for further analysis ., Since the classifications may not always be at the same depth in the haplogroup hierarchy , Figure S1 , we compared the results up to the common level in the tree and accepted the classification if it was in agreement ., In practice , an unassigned sample for the tandem approach is selected for manual , expert analysis ., Experienced personnel examine the haplogroup assignment from the individual classifiers for familiar patterns ., The confidence values from the classifiers may also be analyzed to resolve frequently seen disagreements ., If the ambiguity cannot be resolved at this stage , SNP testing is done to ensure a correct haplogroup label ., The result of the SNP test is then added to the training set to continually improve the classifiers ., For the nearest neighbor heuristic we used the L1-norm distance metric combined with the following rules ., If the sum of allele value differences between a novel sample and one in the training set was zero , it was an exact match and the novel sample was labeled with the matching samples haplogroup ., If the allele values differed by only one or two , and the samples by which it differed were all in the same haplogroup , it was considered a match resulting from a stepwise mutation and again labeled with the matching samples haplogroup ., Otherwise , the sample was left unassigned ., Table 1 shows the average overall performance of the classifiers , including tandem agreement and the nearest neighbor heuristic , for ten iterations of the 5 fold cross-validation on our ground-truth training set ., The support vector machine was the best performing individual classifier with 95% accuracy ., The performance of the Bayesian classifier and the decision trees was very comparable ., The results for the tandem strategy show that of all the samples we attempted to classify , 86% were in agreement , and that almost 99% of those predictions were correct ., Furthermore , the 14% unassignment rate of the tandem approach was much lower than the 26% of the nearest neighbor heuristic ., The average accuracy for each of the classifiers per haplogroup is shown in the top panel of Figure 2 , and the haplogroup frequency of the training data is below it in the bottom panel ., It is clear from the figure that the accuracy of classification for a particular haplogroup is dependent on its frequency in the data ., We also observe that the support vector machines perform the best , particularly in cases where training data for a haplogroup is most sparse ., We believe that more training data from sparse groups , such as A , B , C , D , H , and N would improve the results to similar levels of more well represented haplogroups such as I , J , and R . The classification accuracy under the tandem approach was very high ., Figure 3 shows the performance for each haplogroup when all the classifiers agreed ., While not all of the classifiers agreed in their output in all cases , we observe from the results in Figure 3 and Table 2 that the rate of agreement was very low mostly in haplogroups with low representation in the data ., Again , as we continue to increase the size and diversity of the training set , we expect that the level of agreement in the tandem approach will continue to improve ., In addition to testing the performance of the classifiers on our 15-locus data set , we also tested them on published STR data collected from West , South and East Asian populations 20 , 21 ., The combined public data sets have 1 , 527 samples of 9 loci at DYS394 , DYS388 , DYS389-I , DYS389-II , DYS390 , DYS391 , DYS392 , DYS393 , and DYS439 ., Figure 4 shows the frequencies of Y chromosome haplogroups in this sample ., We performed two types of experiments with this data ., We first looked at performance using the public data both for training and testing with a 5 fold cross-validation ., We then used our ground-truth data set restricted to the 9 applicable loci as training data , and tested the performance on the entire public data set ., As before when testing the classifiers on our ground-truth data , we ran 10 iterations of five fold cross-validation on the public data ., Table 3 gives the averaged results ., In order to provide a meaningful comparison across the two data sets , the table also shows cross-validation results on the 9-locus subset of our data; six of the loci are not shared by both sets and may affect discriminative abilities of the classifiers ., We observe that the classification accuracy between the two data sets is comparable ., Indeed , the cross-validation on the public data has slightly better results ., Figure 5 and Figure 6 show the average per haplogroup classification accuracy for cross-validation on the public data ., Compared with the earlier cross-validation results on our ground-truth data ( Table 1 ) , the 9 Y-STR subset has a much lower performance than the original 15 Y-STR set ., This implies that the 6 excluded markers contribute to a non-negligible increase in performance ., Thus , if the public data set had these additional markers , we expect that its accuracy under cross-validation would also improve ., We tested haplogroup classification for the public STR data using classifiers trained with the 9-locus subset of our data set ., The classification accuracy results are reported in Table 4 ., Figures S2 and S3 show the average per haplogroup performance ., Although the tandem approach still out-performed the nearest neighbor method , the overall performance shows a decrease in accuracy ., We believe the performance is lower for two reasons: as we have already shown , training with 9 versus 15 Y-STRs substantially reduces the accuracy of classification ( an almost 7 point reduction for the tandem approach when contrasting Table 1 with the lower panel of Table 3 ) ; and the origins of the samples in the public data sets are from populations that are not as well represented in our data set ., In this paper we have shown that by using machine learning algorithms and data derived only from a set of Y-linked STRs , it is possible to assign Y chromosome haplogroups to individual samples with a high degree of accuracy ., We note that the number of Y-STRs used has a significant impact on the accuracy of haplogroup classification ., Our classification software provides a single turnkey interface to a tandem of machine learning algorithms ., It is extensible in that other high-performing classification algorithms can be added to it when they are developed ., We have made the software freely available to use for non-commercial purposes and posted it online at http://bcf . arl . arizona . edu/haplo ., Future work could focus on identifying an optimal set of Y-STRs to obtain the highest accuracy of haplogroup classification ., Our preliminary results ( data not shown ) suggest that different Y-STRs are informative for different haplogroups ., Additional work should help to better understand the properties that make different Y-STRs more or less informative for proper haplogroup assignment ., We have assumed in our Bayesian model that the Y-STR loci are statistically independent given the haplogroup ., While we have observed good performance for this model , it most likely does not reflect the true relationship among loci ., As more information about loci linkage becomes available and our ground-truth data set continues to expand , we could relax this assumption and begin to include such dependencies ., Our Bayesian model assumes that Y-STRs are statistically independent conditioned on the haplogroup ., While we observe good performance using this model , this assumption is not realistic given the lack of crossing over among the Y-linked STRs used in our analysis ., On the other hand , Y-STRs mutate independently in a stepwise fashion , which may cause particular Y-STRs to be effectively unlinked on some haplogroup backgrounds ., As more information about linkage becomes available and our ground-truth data set continues to expand , we may be able to include such information to improve our model ., The software system can be effectively used to construct high throughput SNP test panels , particularly in the case of platforms that restrict the number of SNPs accommodated per panel ., Given a corpus of STR data , the classifiers can identify a collection of candidate SNP sites to be placed on the panel to provide maximum coverage over potential haplogroups in a population ., In this way the software provides a cost-effective first step in a multi-level process for deep haplogroup identification by facilitating targeted SNP testing ., In a decision tree classifier , we learn a set of rules for separating samples into hierarchical classification groups according to locus and allele values ., The internal nodes of the tree are comprised of locus tests for specific allele values and the terminal nodes represent haplogroup classification ., The set of tests from the root node in the tree to a terminal node is the classification rule for a haplogroup ., The tree is constructed from a set of training data using the C4 . 5 algorithm 24 , which hierarchically selects loci that best differentiate the training data into haplogroups ., The locus tests are constructed using a measure of information gain , which is based on information entropy 25 ., The entropy of a random variable quantifies its randomness or uncertainty ., In the case of haplogroups , entropy indicates how much diversity there is in the sample set ., Let ng be the number of samples in haplogroup g ., The entropy for G haplogroups over the data set is defined as ( 3 ) where p ( g ) =\u200ang/N is the probability of the gth haplogroup in the data set ., Thus , higher entropy suggests higher diversity and a more uniform frequency of haplogroup representation in the sample set ., Knowing the allele value of a locus may affect the entropy of the data; additional information either does not change or decreases the entropy ., When a particular allele at the ith locus is known , the conditional entropy is given by ( 4 ) where pi ( g|x ) is the probability the gth haplogroup has allele x at locus i ., Let be the number of samples with the latter characteristic and be the total number of samples in the data with allele x at locus i ., Then pi is defined as ( 5 ) We obtain a general conditional entropy for each locus by marginalizing out the allele values ., This is equivalent to computing a weighted average of Equation 4 , where the weights are given by the probability of each allele ., ( 6 ) ( 7 ) ( 8 ) where pi ( x ) is the probability of allele x at the ith locus over all samples in the data ., It is given by ( 9 ) The general conditional entropy in Equation 8 tells us how much Y-STR allelic variation is associated with a given haplogroup ., A lower value indicates the allele values at the locus explain or predict the haplogroup well ., This leads to the concept of information gain ., The difference in variation among haplogroups when a Y-STR allele is both known and unknown is the information gain ., It is a measure of how well a locus explains haplogroup membership ., Formally , it is defined for the ith locus in the data as ( 10 ) The information gain will always be non-negative , since for all loci ., Given the data set , we trained a decision tree by hierarchically computing the information gain for each Y-STR ., A branch in the tree is created from the locus yielding the maximum gain ., The branch is a test created using the selected locus to divide the data set into subsets grouped by haplogroup and ( possibly shared ) allele values ., Tests at lower levels of the tree are constructed from these subsets in a similar fashion ., Once all the samples in a subset are in the same haplogroup , a terminal leaf on the tree is created , which represents a classification ., Figure 7 illustrates this process ., To classify a new sample , we begin at the root and evaluate the locus tests down the tree with its allele values until a terminal node , representing the classified haplogroup , is reached ., The general decision tree approach has some limitations , including overfitting by creating too many branches and locus bias ., The former can be handled by introducing thresholds or other heuristics for the amount of information gain required to create a branch ., The latter is a more fundamental problem of the approach; by definition , the information gain favors Y-STRs taking many different allele values ., We used the PART and J48 implementations 26 of the decision tree algorithm in order to mitigate the effects of some of these limitations ., In the non-parametric Bayesian model , we define a posterior distribution over the haplogroups conditioned on observed allele values ., The posterior is expressed as the normalized product of the data likelihood and model prior ., For a given sample of allele values , the posterior gives a probability for each haplogroup it could belong to ., It is defined as ( 11 ) where c is a normalization constant , and p ( ⋅ ) is the prior probability over the haplogroups ., The likelihood function , , is a measure of how likely it is that haplogroup g generated sample x ., The fundamental assumption of our naive Bayes model is the independence of the Y-STRs X\u200a= ( X1 , … , XL ) , given the haplogroup g ., A number of possible sources of dependency exist that could weaken the validity of this assumption ., For example , Y-STRs are located on the same chromosome and physically linked , which introduces co-inheritance and the possibility of statistical linkage over short time scales ., However , such statistical relationships are not sufficiently understood to be easily incorporated ., Furthermore , attempting to exploit them through direct use of our ground-truth training data is not feasible because the relatively large number of dimensions 15 would require far more data ., In short , the simplifying conditional independence assumption makes using our data tractable ., Interestingly , the accuracy of naive Bayes classifiers is not tightly linked to the validity of this assumption 27 , 28 , which directly affects the accuracy of the posterior computation , but only indirectly affects the ability of the model to distinguish between groups on real data ., In practice , the naive Bayes classier often performs well , and thus we chose to empirically study it for haplogroup identification ., Mathematically , the independence assumption leads to defining the likelihood as a product over each Y-STR density function , ( 12 ) We estimated the density functions fi ( ⋅ ) using histograms constructed from the data ., For each Y-STR and haplogroup , we created a normalized histogram from the training data with bins corresponding to the different allele values the Y-STRs can take ., For the ith locus under haplogroup g , the bins for allele value x are given by ( 13 ) As an example , a set of L densities for a haplogroup and how they are evaluated for a given sample are shown in Figure, 8 . The distribution Equation 11 is defined over all haplogroups , but is not by itself a classifier ., To make a decision , we choose the maximum under the posterior ( MAP ) ( 14 ) This minimizes our risk of an incorrect classification ., A benefit of the generative classifier is the ability to provide alternative classifications and a real probability associated with each decision ., Support vector machines learn a hyperplane with maximal margin of separation between two classes of data samples in a feature space 29 , 30 , 31 ., We trained SVMs for binary haplogroup classification by treating locus alleles for a sample as L-dimensional vectors in Euclidean space and learning a hyperplane to separate them ., A new sample is classified according to which side of the hyperplane its allele values fall on ., We first address the case of deciding between two haplogroups to describe the standard support vector machine approach ., We then introduce a method to combine binary classifiers into a multi-way classifier for all haplogroups using evolutionary evidence for haplogroup relationships ., For a sample xn of locus alleles , consider the task of deciding between two haplogroups with labels {−1 , 1} ., If we assume the locus allele values between the two haplogroups are linearly separable in some feature space , we can use the classification model ( 15 ) where y ( x ) =\u200a0 is a L-dimensional hyperplane separating the two haplogroups; φ ( ⋅ ) is any constant transformation of the allele values into a feature space ., Thus , for the nth sample , the haplogroup is gn\u200a=\u200a1 when y ( xn ) >0 and gn\u200a=\u200a−1 when y ( xn ) <0 ., The goal of training a support vector machine is to find the hyperplane , defined by w , b in Equation 15 , giving the maximum margin of separation between the data points in the two haplogroups , Figure, 9 . The margin is defined as the smallest perpendicular distance between the separating plane and any of the data points in the sample ., By noting that the distance of a sample xn from the hyperplane is |y ( xn ) |/||w|| , and that gny ( xn ) >0 for all samples in the training data , then the maximum margin solution is described by the optimization ( 16 ) However , solving this optimization problem directly is difficult , so we re-formulate it as follows ., Without loss of generality , we can rescale w , b so that the sample ( s ) with allele values closest to the hyperplane satisfy ( 17 ) as in Figure, 9 . Then the optimization problem Equation 16 reduces to maximizing ||w||−1 , which is equivalently re-formulated for convenience as ( 18 ) with the constraint that gn ( wTφ ( xn ) +b ) ≥1 , for all 1≤n≤N ., This can be solved as a quadratic programming problem by introducing Lagrange multipliers an≥0 for each constraint , giving the function ( 19 ) By differentiating L ( ⋅ ) L ( ⋅ ) with respect to w and setting it equal to zero , we see that ( 20 ) Substituting the above into the classification Equation 15 , we obtain ( 21 ) where the kernel function is defined as k ( xn , x ) =\u200aφ ( xn ) Tφ ( x ) ., Therefore , training the model amounts to solving the quadratic programming problem to determine the Lagrange multipliers a and the parameter b ., This is typically done by solving for the dual representation of the problem ., Transforming the problem into its dual shows that the optimization exhibits the Karush-Kuhn-Tucker conditions that ( 22 ) ( 23 ) ( 24 ) Therefore , every sample in the training set will either have its Lagrange multiplier an\u200a=\u200a0 , or gny ( xn ) =\u200a1 ., The samples whose multiplier is zero have no contribution to the sum in Equation 21 , so they do not impact the classification ., The samples that have non-zero multipliers are the support vectors and lie on the maximum margin hyperplanes , as in Figure 9; they define the sparse subset of data used to classify new samples ., A common and effective kernel to use for SVMs is the Gaussian , which has the form ( 25 ) We chose to use this kernel and assume the haplogroups are linearly separable in this transformed space over locus allele values ., In order to make the SVM approach work on data that may not be perfectly separable , we allow for some small amount of the training data to be misclassified ., Thus , rather than having infinite error when incorrect ( zero error when correct ) , we allow some of the data points to be classified on the wrong side of the separating hyperplane ., To accomplish this , we follow the standard treatment of introducing slack variables that act as a penalty with linearly increasing value for the distance from the wrong side 30 , 31 ., A slack variable ξn≥0 is defined for each training sample with ξn\u200a=\u200a0 if the sample is on or inside the correct margin boundary and ξn\u200a=\u200a|gn−y ( xn ) if it is incorrect ., So we now minimize ( 26 ) where ξn are the slack variables , one for each data point , and C>0 weights the significance of the slack variables to the margin in the optimization ., The optimization process is similar to before , but the Lagrange multipliers are now subject to the constraint 0≤an≤C ., As before , the samples whose multipliers are non-zero are the support vectors ., However , if an\u200a=\u200aC , then the sample may lie inside the margin and be either correctly or incorrectly classified , depending on the value of ξn ., Since SVMs train a binary classifier and we have multiple haplogroups to distinguish between , we trained an SVM for each haplogroup in a one-vs-many fashion ., In general , an SVM trained as one-vs-many for a particular haplogroup uses samples in that haplogroup as positive exemplars and samples in other haplogroups we wish to compare against as negative exemplars ., We organized the set of binary classifiers into a hierarchy based on the currently known binary haplogroup lineage 12 , 13 ., At each level of the hierarchy , Figure 10 , the one-vs-many classifiers are trained using only samples with haplogroups at that level , descendant levels , or ancestors; the samples at other branches are not used ., Classification down the tree is accomplished by choosing the SVM result that has a positive classification ., When there is more than one positive classification ( or all negative ) , we choose the result with the closest distance to the support vectors ., If the haplogroup the sample is best associated with is not a leaf node , it is further evaluated down the tree until a leaf is reached ., In order to create a high throughput classification system , all samples of STR data needing haplogroup prediction are batch selected from a database at regular intervals and classified ., Once selected , our tandem classification software predicts the haplogroup for each sample and updates its record in our laboratory information management system ( LIMS ) ., Laboratory technicians and data reviewers can then view the results in a web interface ( Figure S4 ) for the classified batch of samples ., The LIMS displays which samples need to be SNP tested for haplogroup verification ( based on lack of tandem agreement ) ., Once verified , the tested samples are added to the ground-truth set to improve future classifications ., The tandem classification software brings together a collection of algorithms implementing naive Bayes , support vector machines , and decision tree classifiers ., Where available , we used standard implementations of these algorithms that are open and available to the public ., For support vector machines , we used the freely available software package libSVM 32 , which is written in C++ ., We added a customized extension to the library to support multi-class haplogroup prediction as previously described , where the set of trained one-vs-many binary SVM classifiers are organized into a hierarchy that follows Figure 10 ., In addition to providing training and binary classification algorithms , the SVM library provides tools to efficiently iterate over possible constants and kernel parameters using cross-validation in order to find the best set to use ., The decision tree classifiers J48 and PART were used as components of the Weka machine learning software suite 26 ., The software is written in Java and called from our tandem classification software as an external program . | Introduction, Results and Discussion, Materials and Methods | Genetic variation on the non-recombining portion of the Y chromosome contains information about the ancestry of male lineages ., Because of their low rate of mutation , single nucleotide polymorphisms ( SNPs ) are the markers of choice for unambiguously classifying Y chromosomes into related sets of lineages known as haplogroups , which tend to show geographic structure in many parts of the world ., However , performing the large number of SNP genotyping tests needed to properly infer haplogroup status is expensive and time consuming ., A novel alternative for assigning a sampled Y chromosome to a haplogroup is presented here ., We show that by applying modern machine-learning algorithms we can infer with high accuracy the proper Y chromosome haplogroup of a sample by scoring a relatively small number of Y-linked short tandem repeats ( STRs ) ., Learning is based on a diverse ground-truth data set comprising pairs of SNP test results ( haplogroup ) and corresponding STR scores ., We apply several independent machine-learning methods in tandem to learn formal classification functions ., The result is an integrated high-throughput analysis system that automatically classifies large numbers of samples into haplogroups in a cost-effective and accurate manner . | The Y chromosome is passed on from father to son as a nearly identical copy ., Occasionally , small random changes occur in the Y DNA sequences that are passed forward to the next generation ., There are two kinds of changes that may occur , and they both provide vital information for the study of human ancestry ., Of the two kinds , one is a single letter change , and the other is a change in the number of short tandemly repeating sequences ., The single-letter changes can be laborious to test , but they provide information on deep ancestry ., Measuring the number of sequence repeats at multiple places in the genome simultaneously is efficient , and provides information about recent history at a modest cost ., We present the novel approach of training a collection of modern machine-learning algorithms with these sequence repeats to infer the single-letter changes , thus assigning the samples to deep ancestry lineages . | genetics and genomics/population genetics, computer science/applications, evolutionary biology/bioinformatics, computational biology | null |
journal.pcbi.1000175 | 2,008 | Preferentially Quantized Linker DNA Lengths in Saccharomyces cerevisiae | Eukaryotic genomic DNA exists in vivo as a hierarchically compacted protein-DNA complex called chromatin 1 ., In the first level of compaction , 147 bp lengths of DNA are wrapped in 1 3/4 superhelical turns around protein spools , forming nucleosomes 2 ., Consecutive nucleosomes are separated by short stretches of unwrapped “linker” DNA ., Most chromatin in vivo is further folded into shorter , wider fibers , ∼30 nm in diameter ., Despite much effort , the structure of the 30 nm fiber remains unresolved 3 , 4 ., Here we report that an analysis of the relative locations of nucleosomes along the DNA sheds new light on chromatin fiber structure ., The connection arises from the helical symmetry of DNA itself 5–8 ., Each base pair increase in separation between two consecutive nucleosomes moves them apart by 0 . 34 nm along the DNA - a potentially minor change relative to the 30 nm fibers width ., However , because of the 10 . 2–10 . 5 bp per turn helical symmetry of DNA , this 0 . 34 nm translation is coupled to a ∼35° rotation about the DNA helix axis , rotating the second nucleosome to an entirely different location in space , creating an entirely different intrinsic chromatin structure ., In vivo , attractive nucleosome-nucleosome interactions 9 , 10 might overwhelm this intrinsic structure for the chromatin fiber , and impose a particular folded structure that is independent of exact linker DNA lengths ., In that case , changes in the fibers intrinsic structure would be manifested instead as changes in the folded fibers stability 8 ., Because of the high torsional stiffness of DNA and the short lengths of linker DNAs , such changes in stability would be of great energetic significance ., While steps of one or several bp profoundly alter the intrinsic fiber structure , steps of 10–11 n bp ( integer n ) do not: instead , the next nucleosome rotates n complete turns around the DNA helix axis , ending up rotationally near where it began , but translated along the DNA by ∼3 . 4–3 . 7 n nm ., If linker DNA lengths varied randomly about an average value , the resulting intrinsic chromatin structure would be a random flight chain ., But if linker DNA segments instead were equal in length modulo the DNA helical repeat , this would define an intrinsically ordered ( but possibly irregular ) superhelical structure for the chromatin fiber , with the detailed intrinsic structure highly depending on the phase offset d0 ( integer ) for linker DNAs of length 10n+d0 bp ., There are many hints in the literature for a ∼10 bp-periodicity in lengths of linker DNAs 5–8 , 11–13; however , the results are inconsistent ., An early analysis of oligonucleosome DNA lengths suggested that linker DNAs in the yeast S . cerevisiae preferentially occur in lengths of 10n+5 bp , while those in human HeLa and chicken erythrocyte cells have no periodicity 5 ., Analogous studies on rat liver chromatin first did not 14 , but later did 6 , reveal periodic linker DNA lengths , again of the form 10n+5 bp ., Later genome-wide correlation analyses of AA and TT dinucleotides ( which favor particular locations within the nucleosome 15 , 16 ) similarly yielded variable results , suggesting preferences of the form 10n+5 11 or ∼10 . 6n+8 bp for yeast 12 , ∼10 . 6n+8 for Caenorhabditis elegans and Drosophila melanogaster , 12 , and ∼10n+8 for human k562 cells 13 ., These conflicting conclusions of existing studies motivated us to develop two new independent computational methods and new experimental data , to define the probability distribution of linker DNA lengths in yeast ., Our results from both approaches show that linker DNA lengths in yeast are indeed preferentially periodic , implying that the yeast genome encodes an intrinsically ordered three-dimensional structure for its average chromatin fiber ., A well-known characteristic of nucleosome DNA sequences is the ∼10 bp periodicity of key dinucleotide motifs , particularly AA , TT , TA , and GC ., AA/TT/TA steps occur in phase with each other , and out of phase with GC 16–18 ., These facts allow one to test for genomically encoded preferences in linker DNA lengths ., Consider a set of experimentally mapped nucleosome sequences S\u200a=\u200a{s1 , … , si , … sI} , strictly aligned at their dyad ( 2-fold rotational symmetry ) axes ., Extend each aligned sequence in both directions along the genome by roughly one nucleosome length ( Fig 1A , note: positions from 1 to 147 stand for the nucleosome region ) ., Occurrences of AA/TT/TA motifs as a function of position in the flanking regions of S would then exhibit collective patterns that are determined by the distribution of linker DNA lengths around the central ( mapped ) nucleosomes ., If the central nucleosomes were perfectly aligned , and linker lengths were a constant , d0 , then the nucleosomes in the up-/downstream region of S would also be strictly aligned ., A significant periodic signal from AA/TT/TA motifs would then occur at up-/downstream positions dependent upon d0 ( Figure 1B ) ., If instead the ith linker length downstream of sequence si , li , equals 10ni+d0 for some integer ni and a fixed d0 ( 0≤d0<10 ) , then the nucleosomes immediately downstream of S would not be strictly aligned , but would instead be offset by a multiple of 10 bp relative to each other ., In this case , the ∼10 bp periodicity of dinucleotide motif signals would be roughly maintained in the extended regions , but more weakly , since end regions of the adjacent nucleosomal DNA in some sequences would be partially aligned over linker DNA in other sequences ., Alternatively , if linker DNA lengths were random , these dinucleotide motifs would lack any significant periodicity in the extended regions ., We used this approach to test for intrinsically encoded linker DNA length preferences in the yeast genomes ., Our in vivo yeast nucleosome sequence collection ( filtered for nonredundant sequences of length 142–152 bp ) contains 296 sequences ., We focus the analysis on the AA/TT/TA signal because this is the most strongly periodic in aligned nucleosome sequences 16 ., Alignment of these sequences ( Figure 2 ) reveal several striking features ., Sharp signals at positions −1 and 149 , and systematic differences in average AA/TT/TA frequency between the original mapped nucleosomes and the extended regions , may reflect sequence preferences of the micrococcal nuclease 19 which is used to biochemically isolate the nucleosomes , or may reflect sequence preferences intrinsic to nucleosomes and linker regions 20 ., Most importantly , the plot reveals hints of a ∼10 bp periodicity in the extended regions , implying that the yeast genomes intrinsically encode preferentially quantized linker DNA lengths of the form ∼10n+d0 ., The value of d0 can be deduced from the positions of the AA/TT/TA peaks in the extended region ., Assume the AA/TT/TA signal appears periodically at positions 8 , 18 , …79 , … 139 within a nucleosome region 15 , 16 ., If the linker length is 5 bp ( or more generally 10n+5 ) , then the expected positions of AA/TT/TA peaks ( as indexed in Fig . 1B ) in the downstream nucleosome region would be 160 , 170 , … , 231 , … 291 ( or these indices+10n if the linker length is 10n+5 ) ., In accord with this analysis , the AA/TT/TA peaks in Figure 2 are roughly positioned at 10s or 1s in the downstream region ., Therefore we conclude that the preferential linker length value in the yeast data obeys the rule 10n+5 ( d0\u200a=\u200a5 ) ., To test the significance of the observed 10 bp periodicity , we first calculated the Fourier transform of the AA/TT/TA signal in the extended region from position 147+d to 147+d+180 ( Figure 2 ) for a given d ., We chose d to be greater than 10 to avoid the sharp peaks observed at the boundaries of mapped nucleosome sequences ( at positions −1 and 149 in Figure 2 ) that likely owe to the micrococcal nuclease enzyme specificity ., We then varied d from 11 to 20 ., The amplitude spectrum ( square root of spectral power ) averaged over all ds is plotted as a function of period in Figure 3 ( red curve ) ., As a control , we constructed 500 randomly shifted samples of the extended regions by choosing a random di value between 11 and 20 for each sequence si , i\u200a=\u200a1 , … I ( see details in Methods ) ., As another control , we obtained 500 random genomic samples , each sample containing the same number of sequences of the same length ( 180 bp ) randomly selected from the genome ., The mean and 95% percentile of the Fourier transform amplitude at each periodicity value derived from these two sets of control samples are also plotted in Figure 3 ., A significant peak at the 5% level ( i . e . , where the average amplitude from the extended samples with fixed ds ( red curve ) exceeds the 95% percentile line from the randomly shifted samples ( green dashed line ) or random genomic samples ( blue dashed line ) ) is observed at period ∼10 . 2 bp ., Because multiple peaks exist around 10 bp periodicity , we use the total power corresponding to period between 9–11 bp as the test statistic ., The total power at 9–11 bp averaged over different ds was compared with that of the random shifted and random genomic samples ., The resulting p-values are 0 . 008 and 0 respectively , refuting the hypothesis that linker DNA lengths within any 10 bp range are uniformly preferred in the genome ., Instead , these significant ∼10 bp periodicities are consistent with the hypothesis that linker lengths in the yeast genome prefer values of the form ∼10n+d0 , for some constant d0 ., Information about preferred values for d0 is contained in the phase of the corresponding ∼10 bp periodicity peak in the Fourier transform ., In Table 1 we present the location of the Fourier transform amplitude peak ( T* ) around 10 bp periodicity , and the corresponding phase angle in radians , for all ds ., If the experimentally obtained nucleosome sequences were perfectly aligned , and if linker DNA lengths were genuinely 10n+d0 for a constant d0 , then shifting the downstream region leftward by d0 bp will synchronize the extended regions AA/TT/TA motif signal with that in the original mapped nucleosome region ., For example , suppose the true linker length is 15 bp ( i . e . , 10n+d0 with n\u200a=\u200a1 and d0\u200a=\u200a5 ) ., As indexed in Figure 1 , the downstream nucleosome AA/TT/TA peaks will be positioned at 170 , 180 , … 301 ., By shifting the downstream region leftward by 5 bp , we will extract the region comprising basepairs 153 ( =\u200a147+5+1 ) and above ., The AA/TT/TA peaks in the extracted region now are positioned at basepairs 18 , 28 , …149 ( relative to the first basepair of the extracted region ) ., We hence expect the phase of Fourier transform at period ∼10 bp from this extracted region to be close to that from the original mapped nucleosome sequences , which had AA/TT/TA peaks positioned at 8 , 18 , …139 relative to their own first basepairs ., Consistent with this analysis , the phase of the AA/TT/TA signal when d\u200a=\u200a15 ( i . e . , when the extended region begins on the 16th basepair after the end of the mapped nucleosomes ) is 2 . 56 in radians , closest among all ds to 2 . 48 , the phase from the original mapped nucleosomes ., Based on this criterion therefore , we conclude the optimal d0 is 5 bp ., This phase analysis for detecting the preferred quantized linker DNA lengths ( i . e . , the preferred d0 ) assumes that the AA/TT/TA motif maintains the same periodicity in the extended region as in the mapped nucleosomes ., This is true: the periodicity having maximum Fourier amplitude ( T* ) equals 10 . 20 bp for both the mapped nucleosomes and the extended regions ( Figure 3 and Table 1 ) ., Hence this analysis implies that linker DNA lengths in yeast are preferentially quantized , with the form ∼10 . 2n+5 bp ., The amplitude in the extended region however is much lower than in the original core region ., This may suggest that that the linker length distribution is not strictly quantized at ( odd ) multiples of 5s , but rather in a form possessing non-degenerate peaks centered around ( odd ) multiples of 5s ., To test the conclusions of the Fourier analysis described above , and to better define the preferred phase offsets d0 , we developed a duration hidden Markov model ( DHMM , 21 ) and used it to analyze a new collection of DNA sequences of dinucleosomes from yeast which we isolated for this purpose ., Dinucleosomes are two nucleosomes connected by their linker DNA ., The DHMM models the dinucleosomes as an oscillating series of two “hidden” states: a fixed-length ( 147 bp ) nucleosome and a variable length linker DNA ., A technical detail is that , as isolated biochemically , dinucleosomes may come with additional short partial linker DNA segments at either end , or alternatively , may be over-digested so as to have an incomplete nucleosome at either end ., We generalize our DHMM to allow for this possibility ., The algorithm predicts the positioning of two nucleosomes ( complete or partial ) in each sequence , and then uses the predicted results to update parameters in the model that describe the length and sequence preferences of the linker DNA ., In particular , as the algorithm proceeds iteratively , the linker length distribution is updated using the kernel smoothing method ( see details in Methods ) ., We isolated and fully sequenced 335 non-redundant dinucleosomes from yeast , with lengths ranging from 280 to 351 bp ., Some of the dinucleosome sequences were shorter than 2*147 bp , meaning that they have been over-digested on at least one of their two ends ., For such sequences the optimal path is more difficult to predict because of loss of information in either end ., Hence we restricted our analysis to 214 sequences whose lengths are ≥300 bp ., At convergence of the model , the results ( Figure 4A ) confirm the results from the independent Fourier analysis of extended mononucleosome sequences ( Figure 2 and Table 1 , above ) : the linker length distribution function FL ( d ) obeys the rule 10n+d0 with the phase offset d0\u200a=\u200a5 bp , such that the most probable linker lengths ( in the kernel-smoothed distribution ) are around 5 , 15 , 25 , 35 , and 45 bp ., The noise reflected in Figure 4A comprises two chief components: individual major peaks can be slightly offset from 5s; also small peaks arise at seemingly random positions ., This variability in the estimated density curve is not surprising , since we are estimating a distribution in an infinite-dimensional space ., To reduce the dimensionality of the problem , we next consider a parametric approach in which we impose a periodicity on the linker length distribution function , FL ( d ) , while allowing variability around each period ., Such a distribution can be characterized by a mixture of Gaussian distributions with means that are equally spaced by 10 bp , and a common unknown variance ( see Methods ) ., The algorithm proceeds in the same way as the kernel smoothing method above , except that the linker length distribution is estimated using an EM algorithm ., The results at convergence of the model ( Figure 4B ) imply that , if the linker lengths do indeed prefer the form specified in this model , then the optimal d0 value is ∼5 bp ., The estimate of the common standard deviation of the Gaussian components was 1 . 43 , indicating a modest uncertainty of the linker length distribution around the quantized values ., We further generalized the linker length model by treating the period as an unknown parameter and assuming heterogeneity in the variance of Gaussian components ., The resulting maximum likelihood estimator of the period is 9 . 8 bp and the linker length distributions closely resemble those of Figure 4B ( results not shown ) ., Taken together , these results confirm the results of the Fourier analysis and of the DHMM kernel smoothing analyses ., All of these analyses imply that linker DNA lengths in yeast obey the form 10n+d0 with d0 equal to ∼5 bp ., One possible concern in the DHMM analyses is whether the ∼10 bp periodicity in the linker length distribution could somehow arise from the model itself , especially given the ∼10 bp periodicity of motif signals inherent in the nucleosomes ., Two simulation studies tested and disproved this possibility ., One test simulated random sequences based on a product multinomial model with base composition and length distribution identical to that in the true dinucleosome sequences; the second test shuffled the natural dinucleosome sequences while keeping the dinucleotide frequency fixed within each sequence ., The DHMM-kernel procedure was followed exactly as before ., In both simulations , the resulting linker length distribution varied between trials , and the ∼10 bp periodicity disappeared in general ( Figure S1 ) ., The DHMM-mixture method imposes the 10 bp periodicity on the linker length , but the peak positions often moved little away from their initial values of μ as the algorithm proceeded - presumably because , unlike for the real sequences , the randomized sequences lack signals that spur the movement of μ in the true nucleosome sequences ., Thus , the real data are distinguished from the random data in both versions of the DHMM ., We conclude that the linker length patterns deduced by these analyses reflect true signals of nucleosome organization present in the dinucleosome sequences ., To evaluate the robustness of these DHMM analyses to over- or under-digestion of the biochemically isolated nucleosomes and dinucleosomes , we carried out a simulation of the entire combined experiment ., We simulated 2000 nucleosome sequences based on the experimentally obtained yeast nucleosome profile ( a heterogeneous Markov chain model ) ., Both ends of each simulated nucleosome were subjected to a random truncation or addition to the 147 bp-long nucleosome core by up to 3 bp , creating a set of simulated yeast nucleosome sequences having lengths in the range 141–153 bp , slightly greater than the 142–152 bp range of lengths in the real nucleosome sequences ., Similarly , we simulated 2000 dinucleosome sequences , each starting and/or ending with a ( simulated ) nucleosome that was subject to a random truncation or addition of up to 20 bp ., The linker DNAs were simulated using the homogeneous Markov chain model obtained from the yeast dinucleosome data , while the true linker length distribution followed a periodic distribution with peaks at 15 , 30 , …105 ( Figure 5A ) ., The length range of resulting dinucleosome sequences is ∼250–440 bp , which we again filtered to retain lengths greater than or equal to 300 bp ., We followed the same center alignment and model training procedure as for the real data ., The periodic linker length distribution was successfully recovered using both the kernel smoothing and mixture model approaches ( Figure 5B and 5C ) ., Similar results were obtained with a small subset ( 300 ) of the dinucleosome sequences ( Figure 5D ) , where the superior performance of the mixture method on this smaller dataset is evident ( Figure 5C vs . Figure 5B and 5E vs . Figure 5D ) ., In another check , we simulated another 2000 dinucleosome sequences having a uniform distribution for the linker length on 1 , …120 ., The resulting predicted linker length distribution ( Figure 5F ) lacks significant periodicity; peaks formed randomly , and their positions varied from sample to sample ., Classic experimental measurement of the nucleosome repeat length provide several additional checks on the results from the DHMM analyses ., Experiments using gel electrophoresis to analyze the DNAs that result from random partial nuclease digestion of chromatin routinely reveal ladder-like patterns of DNAs fragments , which reflect a repetition of a ( relatively ) discrete sized structural unit comprising a nucleosome plus one average linker DNA length ., The length of DNA in this repeating unit is referred to as the nucleosome repeat length ., Specifically , the nucleosome repeat length may be defined , and measured , as the average length difference in base pairs between DNA fragments containing n+1 and n nucleosomes ., In one test of our analyses , we find that the average length of linker DNA for yeast predicted from the kernel smoothing method is 20 . 2 bp ( 21 bp from the mixture model ) , in good agreement with the experimental value of ∼18 bp for yeast 22 , and in good agreement with subsequent studies suggesting that ∼20 bp may be a more-accurate value than ∼18 bp 3 , 8 ., As a second check on our analyses , we simulated the complete experimental measurement of nucleosome repeat length ., We first simulated the chromatin fiber itself , given our linker length distribution function deduced from the DHMM analysis , then simulated the random partial nuclease digestion , and then finally simulated the gel electrophoresis analysis of the resulting DNA fragments ., The simulated chromatin fibers comprised 50 , 000 nucleosomes with linker DNA lengths distributed as from the mixture model results , i . e . , μ\u200a= ( 5 , 15 , 25 , 35 , 45 ) , σ\u200a=\u200a1 . 43 and η\u200a= ( 0 . 3271 , 0 . 1682 , 0 . 1636 , 0 . 2243 , 0 . 116 ) ( μ was rounded to integers for convenience ) ., The simulated nucleosome chain was then subjected to a simulated nuclease digestion ., To mimic the partial nuclease digestion conditions used experimentally , each linker was subject to zero or one enzyme cut , at a random position and with probability proportional to its length , such that the resulting DNA fragments had a wide range of numbers of nucleosomes , with a mean of 5 nucleosomes ., The simulated gel intensity profile ( Figure 6 ) resembles those observed experimentally ., Thus , the complex linker DNA length distributions deduced in our DHMM analyses are consistent with the experimental observations of ladder patterns in nuclease digestions of chromatin ., Finally , as a third check on our analyses , we used these simulation-derived plots of frequency versus fragment size to recover an apparent nucleosome repeat length ., The average repeat length based on 50 simulations was 168 . 5 bp with standard deviation 1 . 0 , which accurately recovered the true theoretic repeat length for this modeled distribution , 147+21 . 3\u200a=\u200a168 . 3 bp ., We conclude from all of these tests that the complex linker DNA length distribution functions deduced with our DHMM analyses represent true features in the dinucleosome DNA sequences , and that they are compatible with available experimental data on nucleosome repeat lengths ., In this paper , we developed and applied two different methods to investigate the distributions of linker DNA lengths in yeast ., Despite being fully independent , and applied to different kinds of experimental data ( genomic DNA sequences adjacent to experimentally mapped nucleosomes , and , separately , sequences of biochemically isolated dinucleosomes ) , both methods lead to the same conclusion: linker DNA lengths are not described by a uniform distribution , but instead are preferentially quantized , obeying the form 10n+5 bp ., Our results accord with some , but not others , of the previous experimental studies of linker DNA lengths in yeast ., Surprisingly , our Fourier analysis could not detect evidence of periodic higher order structure in the recent genome-wide map of yeast H2A . Z-containing nucleosomes 23 , using either their coarse-grained or fine-grained calls ., Using a nonredundant subset comprising 1617 of their best-mapped nucleosomes ( those which reveal the nucleosomes periodic AA/TT/TA signature 23 our Fourier analysis of dinucleotide frequency in the corresponding extended regions did reveal a ∼10 bp periodicity , with a phase offset d0\u200a=\u200a5 bp , equivalent to that observed with our smaller number of conventionally sequenced yeast nucleosomes; however this periodicity did not pass a test for significance at the 0 . 05 level ., We suspect that the mapping accuracy of that genome-wide nucleosome collection , which includes nucleosome DNA fragments ranging in length from ∼100–190 bp that are sequenced at only one end , may simply be inadequate to reveal the fine-scale structure revealed by analysis of our conventionally mapped and sequenced nucleosomes ., It is possible that our yeast nucleosome collection may be enriched for an especially stable subset of nucleosomes due to sampling bias imposed by nucleosome mapping technology , and thus could reflect a particular chromatin structure that is enriched in such genome regions ., That said , however , our ongoing analysis of more than 50 , 000 newly mapped unique yeast nucleosome sequences ( accounting for ∼67% of the entire genome ) leads to exactly the same conclusions regarding linker DNA lengths in yeast ( unpublished results ) , suggesting at least that this linker length form 10n+5 is representative of much of the yeast genome ., Nevertheless , we note that our present analysis reveals only a single average most probable linker length distribution ., It remains possible that the detailed distribution of linker DNA lengths ( and corresponding intrinsic chromatin fiber structures ) may vary with location throughout the genome ., It is also possible that different species could have different most-probable linker DNA length distributions ., Indeed , our ongoing study suggests that linker DNA in human k562 cells human may preferentially occur at lengths that are quantized at 10s ., This result however is preliminary and requires further investigation ., Several aspects of our findings are significant ., The existence of preferred linker DNA lengths that are constant , modulo the DNA helical repeat , implies an ordered superhelical structure for the average intrinsic chromatin fiber ., The spread of detailed linker DNA lengths around the preferred quantized values ( Figure 4 ) could reflect random disorder about an intrinsically ordered structure; or it could actually reflect the opposite of that , namely , a tendency to improve the local structural order by compensating for inevitable sequence-dependent differences in the intrinsic helical twist of DNA 24 ., The 5 bp phase offset means that , on average , consecutive nucleosomes in the yeast genome tend to start from opposite faces of the DNA double helix ., Our work also introduces two approaches for the analysis of linker DNA lengths in any eukaryote for which the needed experimental data are available ., In the Fourier analysis , an implicit assumption we made is that the nucleosome cores in the extended regions have the same features as the mapped ones , including the periodicity and relative phases of AA , TT , TA , and GC signals ., The justification for this assumption is that these features of nucleosome DNA sequences are thought to reflect the requirement of DNA wrapping in the nucleosome , and to be generic to all nucleosomes 17 , 25 ., The success of the Fourier method highly depends on both the alignment quality and on the extent to which the linker DNA lengths are actually quantized ., A bad alignment tends to degrade the 10 bp periodicity of AA/TT/TA signal in the extended region , just as occurs in the randomly shifted samples ( i . e . a randomly shifted sample can be regarded as resulting from randomly aligned nucleosome sequences ) ., In reality the center alignment is not perfect due to various factors such as sequence specificity of the nuclease which is used to biochemically isolate the nucleosomes ., Hence we believe that the AA/TT/TA periodicity in the extended region based on a “true” alignment would be even stronger than as obtained in Table 1 ., The DHMM provides a general framework for analysis of the linker length distribution function ., The components of the DHMM ( e . g . , the model for the nucleosome sequences or the lengths and sequences of the linker DNA ) are not limited to what have been used in this paper: any probabilistic models for the two states can be readily adapted into this framework ., The legitimacy of the conclusion regarding the linker DNA length distribution , which is drawn based on the DHMM model , depends on the validity of the model assumptions ., Markovian models have proved exceedingly successful in modeling natural DNA or protein sequences in various important problems ., In this paper , we proposed a first-order inhomogeneous Markov chain model for the nucleosome state ., This model is explicitly designed to characterize the sequential dependence of nucleosomal DNAs in the form of dinucleotides ., In addition , it accounts for the variation of signal intensity as a function of positions within the nucleosome region ., The need for representing dinucleotides instead of just mononucleotides was explicitly demonstrated in our earlier study 16 ., Similarly , the distinction of transition probabilities among positions in the nucleosome region is essential in the prediction of nucleosome positioning , given that the dinucleotide signals are known to be periodic 17 , 25 ., As expected from these past studies , our training data show that the transition probabilities are NOT homogeneous at different positions across the nucleosome core region ., The resulting nucleosome model contains a large number of parameters in the transition matrices ( see Methods ) because of this time-dependence ., Nevertheless , from this perspective , over-fitting is not a big concern in this model ., In addition even if this assumption were mis-specified , the trained transition probabilities are still unbiased and consistent estimates of the true parameters ., The only loss incurred would be some asymptotic efficiency of the estimates from a statistical point of view ., One limitation is that the DHMM is a complex machinery , involving many parameters ., Thus we are unable to provide a measure for uncertainty in terms of the entire distribution of linker length , other than the variability around the quantized values quantified by the DHMM-mixture approach ., This remains as an open problem ., We obtained 296 nonredundant 142–152 bp long in vivo nucleosome DNA sequences from yeast as described 18 ., These sequences were mapped to the genome using BLAST 26 ., Dinucleosomes ( experimentally isolated chromatin oligomers containing just two nucleosomes ) were purified from yeast as described 18 , except using less micrococcal nuclease and then gel purifying , cloning , and sequencing protected dinucleosomal DNAs instead of mononucleosomal ., We isolated and fully sequenced 335 non-redundant dinucleosomes , with lengths ranging from 280 to 351 bp ., These were subsequently filtered ( see Results ) to yield 214 sequences whose lengths are ≥300 bp ., We compared the 296 mononucleosome sequences with the 214 dinucleosome sequences that were at least 300 bp , and found only 4 of them were overlapped ., Therefore these two collections can be regarded as two independent sets ., The center of each experimentally mapped nucleosome DNA sequence was treated as the dyad symmetry axis and was indexed as position 74 ., We then extended the genomic DNA sequence on both strands in the 3′ direction for 200 bp ., The resulting extended sequences were aligned according to the center of the mapped nucleosome sequences ( Figures 1A , and 2 ) ., We denote the extended sequence as S\u200a=\u200as1 , …sI ., We sequentially obtained the aligned chunk of DNA of length L0 from position ( 147+d+1 ) to ( 147+d+L0 ) for d\u200a=\u200a11 , … , 20 bp in the downstream region ., d is chosen to be greater than 10 bp to avoid sharp peaks observed at the nucleosome boundaries ., ( Differing values d do not influence the observed periodicities , rather , they lead to slight perturbations of amplitude , because of variation of base composition . We then average the results obtained over the set of d values . ), The average linker DNA length in yeast is ∼20 bp 1 ., We therefore chose L0\u200a=\u200a180 bp such that the extended block roughly covers three full nucleosomes for each sequence ., We further generated 500 randomly shifted samples as follows ., For eac | Introduction, Results, Discussion, Methods | The exact lengths of linker DNAs connecting adjacent nucleosomes specify the intrinsic three-dimensional structures of eukaryotic chromatin fibers ., Some studies suggest that linker DNA lengths preferentially occur at certain quantized values , differing one from another by integral multiples of the DNA helical repeat , ∼10 bp; however , studies in the literature are inconsistent ., Here , we investigate linker DNA length distributions in the yeast Saccharomyces cerevisiae genome , using two novel methods: a Fourier analysis of genomic dinucleotide periodicities adjacent to experimentally mapped nucleosomes and a duration hidden Markov model applied to experimentally defined dinucleosomes ., Both methods reveal that linker DNA lengths in yeast are preferentially periodic at the DNA helical repeat ( ∼10 bp ) , obeying the forms 10n+5 bp ( integer n ) ., This 10 bp periodicity implies an ordered superhelical intrinsic structure for the average chromatin fiber in yeast . | Eukaryotic genomic DNA exists as chromatin , with the DNA wrapped locally into a repeating array of protein–DNA complexes ( “nucleosomes” ) separated by short stretches of unwrapped “linker” DNA ., Nucleosome arrays further compact into ∼30-nm-wide higher-order chromatin structures ., Despite decades of work , there remains no agreement about the structure of the 30 nm fiber , or even if the structure is ordered or random ., The helical symmetry of DNA couples the one-dimensional distribution of nucleosomes along the DNA to an intrinsic three-dimensional structure for the chromatin fiber ., Random linker length distributions imply random three-dimensional intrinsic fiber structures , whereas different possible nonrandom length distributions imply different ordered structures ., Here we use two independent computational methods , with two independent kinds of experimental data , to experimentally define the probability distribution of linker DNA lengths in yeast ., Both methods agree that linker DNA lengths in yeast come in a set of preferentially quantized lengths that differ one from another by ∼10 bp , the DNA helical repeat , with a preferred phase offset of 5 bp ., The preferential quantization of lengths implies that the intrinsic three-dimensional structure for the average chromatin fiber is ordered , not random ., The 5 bp offset implies a particular geometry for this intrinsic structure . | mathematics/statistics, computational biology | null |
journal.pbio.1002096 | 2,015 | A Neural Mechanism for Time-Window Separation Resolves Ambiguity of Adaptive Coding | In many sensory pathways , adaptation serves to adjust neural encoding to the current statistics of the environment and thus enables reliable and invariant representation of relevant aspects within seconds 1–3 ., In order to achieve this , response normalization has been shown to remove the signal mean 1 , 4 , 5 , variance 6–8 , or even higher statistical moments from the representation 9 along the sensory pathway ., From another point of view , this process could be seen as a filter of some statistical moments of the stimulus ., However , although filtering out certain aspects of the stimulus may be desirable for some tasks , it could create ambiguity for others 10 , 11 ., For example , a major function of adaptation is to keep the representation of the stimulus within the dynamic range of the sensory pathway ., Hence , adaptation to the mean intensity usually takes place early on in the periphery , often even within receptor cells or after the first synapse 4 , 12 , 13 ., Consequently , information about the mean intensity should be lost to all later stages of a divergent pathway ., How sensory pathways deal with this problem is still under research ., It has been suggested that multiplexing the information in different aspects of the response statistics may solve the problem 10 , but the identification of mechanisms to read out such a code has been difficult 14 ., When possible , adaptation can be placed after divergence of the processing of different aspects 15 , 16 or more generally spread across larger populations 17 , 18 , but because of limited dynamic range or processing capacity , in many cases this is not a viable solution ., The conflict is particularly prominent in the auditory system , in which differences in the mean sound pressure level between the two ears are used to localize a sound in the horizontal plane 19 , while a level-invariant representation of the amplitude modulations of the sound is an important prerequisite for recognition of sound identity 20 ., Across a wide range of species , adaptation creates level-invariant representations of amplitude modulations 4 , 5 , 7 , 13 , most likely to accommodate for the large range of mean intensities at which even the same sound can be encountered ., However , if level-invariance is created by adaptation already in the periphery , central neurons are not able to evaluate inter-aural level differences ( ILDs ) , since mean intensity is already removed from the neural responses ., The auditory system of the grasshopper is perfectly suited to study this potential conflict between peripheral adaptation and central processing of ILDs , since only two features are of major behavioral importance in this system: ( 1 ) temporal pattern of the signal that serves both males and females to detect and identify a potential mate 21 and ( 2 ) the ILD , on which the animals rely to localize the sound source in order to approach the potential mate 22 ., The ears of grasshoppers are located laterally in the first abdominal segment ( Fig . 1A ) ., Receptor neurons transduce sound and encode information about the stimulus pattern in action potential frequency 23 , 24 ., The receptor axons enter the metathoracic ganglion , where they synapse on a population of local neurons ( Fig . 1B , 25 ) ., In this ganglion , information about the pattern and the directionality of the sound is separated into two channels , represented by different ascending neurons 21 , 26 ., Both channels make use of the same peripheral input from both ears combined in central , ascending neurons ., Receptors do not synapse directly onto ascending neurons , but on local neurons only ( Fig . 1B ) ., In the case of pattern coding , summation over the peripheral responses from both sides increases the signal-to-noise ratio but leads to a loss of directional information ., For directionality , the system evaluates ILD , whereas inter-aural time differences are much too small to be evaluated in grasshoppers ( differences of 5–6 μs at most; 27 ) ., In order to evaluate ILDs , the grasshopper ear works as a pressure gradient receiver 28 ., In addition , the differences between the peripheral inputs from the two ears are enhanced by contralateral inhibition , emphasizing the directional tuning ., The ascending neuron AN2 in locusts is thought to code for the direction of the stimulus 25 ., We have previously quantified adaptation at different stages of the metathoracic network and discovered that adaptation takes place at all levels of the pathway 29 ., Here we explore how peripheral adaptation influences the central representation of pattern and ILDs and how the network deals with potentially conflicting requirements on adaptation for pattern and ILD coding ., In a first step , we experimentally explored at which stage in the metathoracic network of locusts ( Locusta migratoria ) level invariant coding is achieved ., Since we had previously observed strong negative feedback in a central direction coding neuron 29 , we next tested whether this central mechanism solves the conflict of adaptation on pattern coding and ILD representation ., Finally , we used a model based on experimental data from all three stages of the pathway to generate qualitative predictions for behavioral experiments and tested these on male grasshoppers of the species Chorthippus biguttulus ., In grasshoppers the peripheral auditory system is located in the metathoracic ganglion ( Fig . 1A , B ) and exhibits three layers of neurons ., Sixty to eighty receptor neurons in the ear encode the sound envelope by action potentials 24 ., Receptor neurons form synapses with local interneurons , which pass information onto ascending neurons ( Fig . 1B ) ., We hypothesized that neural adaptation should take place before central integration of both sides in ascending neurons in order to be beneficial for pattern coding ., However , since local neurons provide the input to ascending neurons of both the pattern and the direction-coding pathway ( Fig . 1B ) , such peripheral adaptation could be detrimental for coding of the direction of a stimulus ., In an initial series of experiments we characterized the strength and effect of adaptation at the first two levels of the peripheral system of grasshoppers: receptors and local interneurons ., We first tested whether the firing rate of local interneurons in response to an ongoing , amplitude-modulated ( AM ) stimulus is independent of the mean intensity of the stimulus ., We recorded intracellularly from a local interneuron ( TN1 ) , while presenting the same Gaussian white-noise AM stimuli ( cutoff 100 Hz ) at different mean levels ., The upper panel of Fig . 1C shows two responses elicited by such a stimulus presented at two different mean levels ., Except for the initially different responses during the first 50 ms ( middle panel in Fig . 1C ) the firing rate of TN1 was very similar , although both stimuli differed by 20 dB in mean level ., We also compared the responses of six different recordings of TN1s to AM stimuli pairwise for the two mean levels over time ., The normalized difference between the responses on average dropped to below 15% after 30 ms and then remained low for the entire presentation ( Fig . 1C , bottom panel ) ., This demonstrated that at the first steps of processing in the auditory pathway of grasshoppers adaptation to mean sound level enabled intensity-invariant responses already at the input to both the pattern and localization pathway ., In order to quantify level invariance in local interneurons more thoroughly , we tested level-response curves either in silence or during presentation of an adapting background stimulus at different mean intensities ( Fig . 1D ) ., If adaptation really results in level invariance , a change of the background level is expected to produce a compensatory shift of the level-response 30 ., Indeed , local neurons responded to the level of the test pulses relative to the background instead of responding to the absolute intensity ( Fig . 1E ) ., Fig . 1F shows an example of a TN1 that shifts its response curve along the intensity axis for different background levels ( dotted vertical lines ) ., A full compensation of the change in mean level by adaptation would correspond to a complete level invariance of responses , indicated by points along the identity line in the left panel of Fig . 1G ( slope of 1 ) ., The shift of the response curves of TN1 was highly correlated with the background levels for the entire range tested ( r = 0 . 95 , p < 0 . 0001 ) , and the slope of 0 . 83 ± 0 . 18 ( 95% confidence interval ) indicated that adaptation compensated for most of the change in signal mean ( Fig . 1G , left panel ) ., Thus , the TN1 exhibited almost complete adaptation to the background ., In order to test how much of this level invariance is inherited from receptor neurons , we performed the same set of experiments while recording intracellularly from seven receptor neurons ., Although response curves of receptors were shifted after the presentation of different background levels , the population data of all receptors ( Fig . 1G right panel ) did reveal less compensation of the presented background intensity by adaptation than local neurons ., The slope of the relationship between the intensity of the adapting background and the shift of the response was 0 . 67 ± 0 . 08 ( linear fit , ± 95% confidence interval; correlation: r = 0 . 94 , p < 0 . 0001 ) , and thus below a slope of one that would be expected for complete level invariance ( dashed line , Fig . 1G right ) ., In summary , nearly complete level invariance is already achieved after the first synapse of the grasshopper auditory pathway , thus removing most of the available information about absolute level at each side of the animal before the separation of channels for parallel processing ., The observed intensity invariant representation of amplitude modulations is likely beneficial for the recognition of song envelopes whose absolute and mean amplitudes depend on the distance between sender and receiver ., However , the auditory periphery does not only feed into the pattern processing circuits , but is also used to determine the direction of a sound source ( Fig . 1B ) , for which grasshoppers mainly depend on ILDs 31 ., Removing available information about absolute level on each side separately also removes information about ILDs ., We next asked how the auditory system of the grasshopper solves this conflict ., Since adaptation evolves over time , one potential algorithm would be to restrict reading of ILD information to the stimulus onset ., In the grasshopper , direction is encoded in two pairs of ascending interneurons that each receives excitatory input from one side and inhibitory input from the other ., In a previous study , we had observed a strong intrinsic activity-dependent adaptation current in one of these neurons ( AN2 ) 29 ., When current was injected into the AN2 , spike frequency quickly dropped down to very low levels ( Fig . 2A ) and often completely disappeared at higher current levels ., In response to sound stimuli the AN2 displayed an even stronger reduction in firing ( Fig . 2B ) ., Since an intrinsic adaptation mechanism restricts firing mostly to the onset of the stimulus that still contains information on absolute sound levels , we hypothesized that intrinsic adaptation could enable the coding of direction despite the observed peripheral adaptation ., To test our hypothesis and to investigate the effect of central adaptation in ascending neurons on direction coding , we simulated the direction-coding pathway of the grasshopper in a network model ., The network consisted of an ipsi- and a contra-lateral population of peripheral neurons that project onto two central , integrating neurons simulated by exponential integrate-and-fire neurons ., The peripheral populations were fitted to match the response curves in the experimental data and the observed dynamics of peripheral adaptation 29 ., One of the two central , direction-coding neurons was excited by sound from the left and inhibited by sound from the right , the other , vice versa ., Parameters for the central neurons were fitted to match the observed spiking responses of AN2 when stimulated with current stimuli ., In order to test for the consequences of intrinsic adaptation currents in the central neuron , we ran the model in two versions: with and without an adaptation term in the central neuron ( Fig . 2B , C ) ., The model version without central adaptation failed to reproduce the experimentally observed responses to acoustic stimuli ( Fig . 2B ) ., Due to peripheral adaptation , the model showed a marked decrease in spike rate but also displayed a sustained response well above zero ., Adding an adaptation current to the central neurons of the model that reproduced the experimental current-injection data resulted in a good match of the firing-rate response of the model to acoustic stimulation with the experimental data ( Fig . 2C ) ., We next tested the performance of our two model versions for the encoding of ILD ., Responses of the two model neurons to an artificial grasshopper song elicited by presentation of three direction-dependent level differences are presented in Fig . 3A ., Initially , for the first sound pulses , the ipsi-laterally excited neuron ( upper trace ) responded with a higher rate than the contra-laterally excited neuron ( middle trace ) ., This was in accordance with the stimulus being louder ipsi-laterally and softer contra-laterally , and this difference in firing rate of the two central neurons potentially encoded the direction of the sound source ., However , as the periphery adapted with time and generated a level invariant envelope representation , the responses of the two neurons become very similar to each other and invariant to the direction ( Fig . 3A , lower panel ) ., The addition of a central adaptation current did not influence the responses to the initial sound pulses of the artificial song in the model ( Fig . 3B ) ., However , for the following sound pulses , the additional central adaptation current strongly reduced the activity in the model AN2 and abolished further representation of the song pattern ( Fig . 3B ) ., Based on the single level example shown in Fig . 3A and Fig . 3B , the grasshopper could discriminate between different directions mostly by a comparison of the onset responses of the two neurons ., However , such discrimination needs to be reliable for a much larger range of different sound levels ., We therefore tested the discrimination performance of our model with 33 different sound levels and ten sound directions resulting in different ILDs ., We then tried to determine the direction of the sound source from the difference of the spiking response of the two model neurons averaged over the entire stimulus ., In the model version without central adaptation the direction of sound was reliably detected only for ILDs larger than 6 dB ( Fig . 3C , left panel ) ., However , the classification success deteriorated at lower ILDs , for which songs often were either classified as coming from the front ( ILD = 0 ) or from a position further to the side of the animal compared to the original ILD ( Fig . 3C left ) ., Therefore , the output of the network did not reliably predict the actual direction if the entire song was taken into account ., Only 58 . 3% of responses were classified within ±1 dB of the original ILD and the mutual information in the confusion matrix in Fig . 3C left panel is 1 . 03 bits ( maximal possible: 3 . 32 ) ., However , more accurate information about sound direction is available by taking only responses to the first syllable into account: performance was much better in this case ( Fig . 3C right panel ) ., These modeling results demonstrate again that adaptation to the mean sound level at the very periphery potentially poses a problem for the localization of a sound source despite the initially unadapted responses that convey information about sound direction ., Since the peripheral pathway of the grasshopper does create level invariance rapidly within about 50 ms ( Fig . 1C and D ) , the question arises how the animals nevertheless successfully locate the songs of potential mates 22 ., When we tested the ability to predict song direction with the second model version ( Fig . 3B ) , the additional , dynamic adaptation current enabled a high ability to discriminate between sound directions ( Fig . 3D ) ., Now , the performance of classification becomes more independent of sound level and thus more invariant for overall levels of sound ., Both correct assignments and mutual information in the classification matrix ( Fig . 3D ) increased by about 32% ( correct assignments within ILD ± 1 dB: 76 . 7% versus 58 . 3% , mutual information: 1 . 36 versus 1 . 03 bit ) ., The central adaptation current suppressed the response to the later sound pulses that do not carry directionality information ., Therefore , the response to the first sound pulses dominated the overall prediction of the model , offering a solution for the conflict posed by peripheral adaptation ., The right panels in Fig . 3C and Fig . 3D demonstrate that this gain was not achieved simply by restricting responses to the first syllable because discrimination performance was better when the modeled response was integrated over the entire song than when only the first syllable was taken into account ( 76 . 7% correct within ILD ± 1 dB versus 65 . 1% correct , 1 . 36 bit versus 1 . 15 ) ., Thus , the adaptation current in AN2 not only cuts off responses when they become uninformative ., The time course of the adaptation enables weighing of responses over time according to how informative they are ., To investigate the effect of taking the weighted difference of responses rather than simply cutting off AN2 responses , we plotted the response difference between ipsi- and contralateral AN2 model neurons at different mean intensities and ILDs ( Fig . 4A and Fig . 4B ) ., Reliable coding of direction would correspond to invariant responses independent of mean sound levels ., For both model variants , however , responses to different ILDs depended on the mean level when only responses to the first syllable were taken into account ., Response differences were highest at intermediate levels and fell off slightly at lower and higher intensities ., Since the classification matrices in Fig . 3C and Fig . 3D were computed for the whole range of mean levels , this dependence on the mean level hindered correct classification of direction ., Taking the entire song into account only worsened the classification performance of the model without central adaptation because response differences at the same ILD became even more level dependent ( Fig . 4A , lower panel ) ., However , adding the adaptation current to the model AN2 made response differences more independent of the mean sound level , as required for an invariant coding of sound direction ( Fig . 4B ) ., We hypothesized that a match of the adaptation time constant of the intrinsic adaptation in AN2 with the peripheral adaptation dynamics was crucial for this improved invariance ., When we tested our model with different time constants for central adaptation , the original value obtained by fitting the time course to the current injection data ( Fig . 2A ) indeed yielded the best classification results ( Fig . 4C ) and highest information transfer about ILD ( Fig . 4D ) ., Thus , we found evidence that the adaptation dynamics in the central direction coding neuron AN2 are matched to the time course of information decay about absolute levels in the periphery and may thus enable level-independent direction discrimination behavior in male grasshoppers ., Our experimental data and the simulations indicated that peripheral adaptation to the mean sound level hampered the discrimination of direction by central neurons ., The potential solution for this conflict we found was to weigh responses over time according to the availability of information about sound direction , by means of an intrinsic adaptation current in the readout neuron ., From this analysis , specific predictions followed for behavioral experiments in which the ability of grasshoppers to localize sound can be tested ., We used the following setup to apply two behavioral paradigms ., A male grasshopper is stimulated simultaneously via two loudspeakers ( Fig . 5A ) ., Whenever the male produces a calling song , we “respond” with a playback of a female song , with slightly different intensities from the two speakers ( ILDs ) ., All stimuli are presented at intensities close to behavioral threshold in order to avoid the effect of sound traveling from the right speaker to the left ear and vice versa 22 ., If the male is able to detect which speaker broadcasts the higher sound level , he will reliably turn towards that direction ., Very small level differences suffice for a correct localization in the normal stimulus situation 22 ., We used short stimuli ( 340 ms ) to establish an open loop situation ., The males start to turn only after approximately 500 ms ( see 32 ) ., Hence , the stimulus was completed before the behavioral response started , giving time for full analysis of the song ., Fig . 5B shows the turning responses to these short songs played at different ILDs ( pattern P1 from Fig . 5A ) ., At an ILD of 1 dB the animals already showed a high performance of more than 80% turns to the correct side ., However , if males could not resolve the level difference between the speakers , i . e . , they perceived the sound as coming from the front , they turned randomly to either side , and in addition , they tended to jump forward ( 22 , 32 , see points at 0 dB level difference in Fig . 5B; here approximately 30% turns and 70% forward jumps occurred ) ., At ILDs of 2 dB the number of correct turns was already maximal and indicated that male grasshoppers lateralized the sound source very precisely ., Our model predicted that if an unstructured adapting stimulus was presented mono-laterally just before the onset of the song ( Fig . 5A , stimulus P2 ) the lateralization should be biased towards the other side ., In this case , the periphery on the adaptor side had become less sensitive and the relative strength of the inputs from both sides to the central neurons sensitive for direction should be shifted ., Therefore , the prediction was that males should turn away from the sound source with the adapting condition P2 and should turn towards the side without adaptor ( P1 in Fig . 5A ) ., Although overall responsiveness to the pattern was lower ( Fig . 5F ) , the experimental results nicely exhibited the trend predicted by the model ., Even when the song with the preceding adaptation was 4 dB louder , the animals still turned preferentially to the other side ( Fig . 5C ) ., Only at a difference of 6 dB , this trend reversed , and the animals correctly turned towards the louder side ., A second prediction of the model was derived from the observation that the central interneuron AN2 responded mainly to the onset of sound patterns ( Fig . 2 ) ., A stimulus that is slowly ramped up from sub-threshold levels should still be recognizable for males ., If the directional response towards a sound source was based only on the onset , however , as the stimulus becomes louder , the periphery is subject to ongoing adaptation , which would strongly reduce directional information ( Fig . 3A ) ., Therefore , males were expected to show a reduced lateralization performance for such ramped up stimuli ., We chose to play back short grasshopper songs that were either ramped up or down in sound level ( Fig . 5D inset ) ., The latter song type should be easier to lateralize for the animals , due to its onset at supra-threshold sound levels ., This was indeed the case ., At 2 dB ILD the animals reached an 82 . 7% performance for the downward-modulated songs , whereas the performance was only 55 . 3% for the upward modulation ( Fig . 5D , difference highly significant , p < 0 . 001 ) ., The result of this test cannot be explained by the males reacting less to the ramped female songs ( Fig . 5F ) ., In addition , when the upward sweep was presented , the animals produced significantly more forward jumps: 37 . 3% forward jumps in the upward case at 2 dB ILD compared to only 9 . 8% for the downward modulation ( Fig . 5E ) ., This result indicated that with the upward sweep , the tested males could not resolve an ILD of 2 dB and , hence , often classified this stimulus as coming from the front in spite of the 2 dB difference ., These results and the results with adapting stimulus ( Fig . 5C ) further confirmed the prediction of our model and strongly suggested that the animals used the onset of a sound pattern as the most reliable information about the direction of a sound source ., In auditory systems of insects , as well as vertebrates , information from both ears is initially processed independently for each side and then combined , for both pattern- and direction-encoding neurons ( see below and Fig . 6A ) ., Adaptation processes generating an intensity invariant representation already in the periphery ( receptors and local neurons in the grasshopper ) are optimal for pattern encoding independent of sound direction 33 ., However , peripheral neurons should not adapt at all for direction encoding in order to preserve information about absolute sound levels at each side ., Furthermore , because of ubiquitous noise in neural responses , response-level curves need to be steep to ensure sufficient discrimination of sound levels ., This requirement implies narrow response curves because of the limited response range of neuronal firing rates ( Fig . 1F ) ., For pattern coding , this is not a problem as long as the response curves are shifted by adaptation to the mean signal intensity ( Fig . 6B , left panel ) ., However , for optimal encoding of direction , monotonically increasing response curves covering a broad intensity range that do not adapt would be optimal ( Fig . 6B , right panel ) ., These arguments are formalized in S2 Text , and clearly demonstrate the conflict between pattern and direction encoding regarding peripheral adaptation and response curve shapes in the steady-state ., The solution to this general problem we present here emphasizes onset transients that still contain directional information before the system is completely adapted ., Another possible solution to the conflict between AM representation and directional coding addressed here would be to rely on two separate populations of peripheral neurons , one dedicated for coding of ILD and one for pattern representation ., As for the receptors , this can be ruled out in grasshoppers ., All recorded receptors shifted their response curve via adaptation ( Fig . 1G ) , thereby removing a good part of the centrally available information about direction ., For the second neuron in the pathway , the local neuron , we cannot be sure whether another neuron exists that does not add to the receptor adaptation , but based on the current knowledge of the circuitry , this seems unlikely 25 ., At this point , the exact wiring pattern of the metathoracic auditory network is only partly known ., We know , however , that the local neuron TN1 receives direct input from receptor neurons 34 ., TN1 itself acts in an inhibitory manner and is the major candidate for the subtractive input current observed in central direction sensitive neurons 25 , 35 ., In addition , the results of the behavioral experiments with the forward masking adaptor ( Fig . 5C ) strongly indicated a major contribution of peripheral adaptation on directionality coding ., As our data and simulations suggest , a separation of pattern and direction encoding in time is a simple , yet powerful , solution to the ambiguity problem introduced by neural adaptation ., In particular , a match of peripheral and central time constants is crucial for this process ., In addition , this mechanism critically relies on the relation between typical timescales of the signals that need to be processed and the adaptation time constants ., From a signal processing point of view , adaptation acts as a high-pass filter and thus does not affect the coding of fast amplitude modulations 30 , 36 ., In the case of the grasshopper song , this applies to the fast amplitude modulations of the stimulus pattern , but also to the sudden availability of directional information at the onset of the stimulus ., Multiple timescales of adaptation and temporal coding in the same pathways have been observed in various systems 36–38 , even scaling of adaptation dynamics with the frequency of rapid changes of signal 10 ., Potentially , some of the variety of timescales found in other systems reflects the dynamics of relevant signals and the need for a temporal separation of information streams , as presented in the current work ., Limiting direction discrimination to the onset of a stimulus comes with two major consequences: the inability to trace moving stimuli and the need for a memory trace of the directional information ., For grasshoppers , the most relevant sound to be localized are sounds of conspecifics , i . e . , of potential mates 31 , 32 ., While the male waits for the female to respond , it sits still , because singing and moving are mutually exclusive—for both actions the legs are used 39 ., The memory trace could be placed anywhere between the ascending , directional neuron ( e . g . , AN2 ) and the movement apparatus ., The directionality could even be stored mechanically in the legs of the grasshoppers 40 while the insect is still evaluating the temporal pattern that eventually triggers the movement ., Because of the larger inter-aural distance , mammals and birds can rely on ILD and inter-aural time differences ( ITD ) 19 ., Neural adaptation and perceptual shifts of perceived direction have been reported for both ITD and ILD coding in mammals 41 , 42 ., In birds , peripheral adaptation has been shown to be detrimental for spike timing precision in the ITD coding pathway 43 ., For high-frequency sound , many mammals rely almost entirely on ILD for horizontal localization ., The mammalian auditory system relevant to ILD coding shares some of the central features of what has been described here ., Peripheral adaptation at similar timescales as in the receptor neurons of the locust has been revealed in recordings from the auditory nerve of guinea pigs 44 , 45 , removing much inter-aural level difference over time 13 ., At later stages of the pathway , neurons have been shown to respond invariantly to mean intensity after a short adaptation period 5 ., For localization , information from both ears is combined centrally , in neurons of the lateral superior olive ( LSO ) ., Similarly to the system presented here , this is done subtractively via excitation and inhibition 46–48 ., Behavioral tests using large ILDs in headphone experiments revealed some adaptation to ILDs 41 , a test very similar to our monaural forward masking experiments ( Fig . 5C ) ., The inputs to central neurons in the LSO display shallow response curve with large dynamic ranges 46 , as predicted by our information theoretic calculations ( Fig . 6B ) ., Neurons in the LSO and higher brain centers have been shown to code ILDs invariant of mean level 48 ., Strikingly , central LSO neurons project back to their inputs via recurrent , presynaptic inhibition 49 , which would have a similar effect as the intrinsic , output-driven adaptation described here—both are implementations of a negative feedback loop ., Thus , the similarity of the task , to process intensity differences between the two ears while accomplishing intensity variance for pattern recognition at the same time , could have led to different implementations of the same algorithm in such distantly related animal groups as insects and mammals ., All electrophysiology was performed in vivo in L . migratoria ., For receptor recordings ( n = 7 | Introduction, Results, Discussion, Materials and Methods | The senses of animals are confronted with changing environments and different contexts ., Neural adaptation is one important tool to adjust sensitivity to varying intensity ranges ., For instance , in a quiet night outdoors , our hearing is more sensitive than when we are confronted with the plurality of sounds in a large city during the day ., However , adaptation also removes available information on absolute sound levels and may thus cause ambiguity ., Experimental data on the trade-off between benefits and loss through adaptation is scarce and very few mechanisms have been proposed to resolve it ., We present an example where adaptation is beneficial for one task—namely , the reliable encoding of the pattern of an acoustic signal—but detrimental for another—the localization of the same acoustic stimulus ., With a combination of neurophysiological data , modeling , and behavioral tests , we show that adaptation in the periphery of the auditory pathway of grasshoppers enables intensity-invariant coding of amplitude modulations , but at the same time , degrades information available for sound localization ., We demonstrate how focusing the response of localization neurons to the onset of relevant signals separates processing of localization and pattern information temporally ., In this way , the ambiguity of adaptive coding can be circumvented and both absolute and relative levels can be processed using the same set of peripheral neurons . | Smell , vision , hearing—virtually all of our senses adapt their sensitivity to cope with the varying environment ., Adaptation removes information about absolute stimulus intensity available to the brain , as this information is usually of little relevance for sensory representation ., For some tasks , however , knowledge of absolute stimulus intensities is essential ., How sensory pathways cope with this conflict remains an open question ., We addressed this question in the grasshopper auditory system , in which comparison of absolute intensities of conspecific calls at both ears is crucial for mate localization ., We recorded activity from three levels in the auditory pathway , showing that adaptation in the peripheral auditory system indeed removes information about absolute intensities ., We discovered that strong negative feedback restricts coding of sound direction in the central auditory system to the very beginning of a stimulus , when peripheral adaptation has not yet acted ., By using a computational model , we show that this central mechanism enables localization of the sound source over a wide range of stimulus intensities and that its time course is well matched to the time course of peripheral adaptation ., In a final step , we confirmed predictions from our model in behavioral experiments on sound localization . | null | Neuronal data, computational modeling, and behavioral experiments reveal how the conflict between sensory adaptation and sound localization is resolved in the grasshopper auditory system, allowing processing of both absolute and relative sound levels. |
journal.pcbi.1000957 | 2,010 | Exploring the Universe of Protein Structures beyond the Protein Data Bank | The total number of distinct protein folds which have been experimentally solved is very small compared to the amount of genome-wide protein sequences 1 , 2 ., Indeed , folds are evolutionarily more conserved than sequences and the same fold can house proteins performing different biological functions 3 , 4 ., Thus a fundamental question concerns the extension of the library of protein folds: are the observed structures a small fraction of the whole fold universe ?, If so , then is it because evolution has not yet run enough to explore it or rather because a selection principle is on which has slowed down/stopped the search for alternatives ?, Addressing these issues on the basis of the principles of physics and chemistry is a question of fundamental importance , currently at the center of intense investigation ., Several properties of the folding process have been shown to depend more on the fold topology than on the specificity of the aminoacids 5–10 ., For a few proteins , native backbone geometries were shown to be closely mimicked by local energy minima of poly-alanine chains 11 ., More recently , a unified approach to the origin of protein folds was proposed in which the inherent anisotropy of a chain molecule , the geometrical and energetic constraints placed by hydrogen bonds , steric hindrance and hydrophobicity yield a free energy landscape with minima resembling protein structures 12–14 ., One of the predictions is that a limited library of folds exists ., Along the same lines , based on a coarse grained model , Zhang et al proposed 15 that there is a one-to-one correspondence between the Protein Data Bank ( PDB ) library and the structures that one can obtain with a homopolymer from the requirement of “having compact arrangements of hydrogen-bonded , secondary structure elements and nothing more” 15 ., A different scenario has been proposed in ref ., 16 where , by using structure prediction method based on an idealized secondary structure lattice representation they argued that the space of possible folds might be larger than the space of natural folds ., Recent advances in supercomputing power and sampling methods 17 , 18 allow us addressing these issues by accurate atomistic simulations ., We here describe the results of a molecular dynamics simulation of a 60 amino acids polypeptide chain performed with an accurate all-atom interaction potential and a setup specifically designed in order to extensively explore the configuration space ., The length of 60 was chosen because it represents the limit of what can be simulated with our computational resources ., Natural proteins are on average much longer than 60 amino acid , but several autonomously folded domains of this size exist 19 , making the comparison between simulation and nature meaningful ., In the simulation we visit practically all the folds observed in nature for proteins of comparable length ., However , at variance with what found in 15 , we find that natural folds are only a small fraction of the structures that are explored ., Many of the structures found in our simulation resemble real proteins ( in terms of secondary content , stability and compactness ) but have not been observed in nature ., This finding immediately rises a question on the nature and meaning of these novel folds: why are they not exploited in real proteins ?, Do natural folds have something “special” or have they simply been selected randomly ?, By using a state-of-the art enhanced sampling technique 18 , we simulate a 60 amino acid polyvaline ( VAL60 ) described by an all-atom potential energy function 20 as explained in Methods ., This allows generating , in of simulation , structures characterized by a significant secondary content and a small radius of gyration ., A movie with a short part of the trajectory ( ) is available as Video S1 ., It shows the exploration proceeds mostly by local reorganization of secondary structure elements ., From time to time the system unfolds completely , exploring a totally independent topology ., A selection of the 30 , 000 structures is represented in Fig . 1-a and a repository , with their all-atom configuration , is available at http://dx . doi . org/10 . 5061/dryad . 1922 ., By steepest descent optimization ( see Methods ) we verified that even if these structures have been obtained with an enhanced sampling technique , they closely correspond to local minima of the potential energy surface of VAL60 ., Consistently with Ref ., 11 , they also correspond closely to local minima of the potential energy surface of polyalanine ( ALA60 ) ( see Methods ) ., Even though these structures correspond to local minima , one still wonders if their structural quality is good and if they resemble real proteins ., In order to address this issue , we monitored several structural quantities on our dataset ., In Fig . 2-a we show the Ramachandran plot of the VAL60 structures ., One can see that the dihedrals populate the allowed regions ., The relative height of the various peaks is determined by the probability to observe the different secondary structural elements and the random coil in the full dataset ., The “stereochemical quality” of the VAL60 set was also assessed using PROCHECK 21 ., This program provides an overall quality measure , called G-factor , which takes into account dihedrals , bond lengths and angles , as compared with stereochemical parameters derived from well-refined , high-resolution structures ., If the G-factor is higher than −1 . 0 the structure is considered to be “normal” ., In Fig . 2-b the G-factor distribution is shown for the VAL60 ., For a comparison , we computed the same distribution also for the structures of length smaller than 75 amino acids belonging to the CATH database 19 ., We also used PROCHECK to estimate the average hydrogen bond energy ., The distributions of this quantity for VAL60 and CATH is shown in Fig . 2-c and compared ( dash line ) with its ideal mean and standard deviation 21 ., For the VAL60 set the G-factor and the H-bond energy , though not as good as for CATH , are in accordance with what is expected for realistic proteins ., Lastly , in order to check if medium size structures generated by our sampling procedure are representative of the PDB , the VAL60 structures were fragmented in small 5 amino acids long structures and were compared by backbone RMSD 22 to all the fragments of the same length found in CATH ., The minimum RMSD value was obtained for each small fragment ., The distribution of this quantity is shown in Fig . 2-d ., It is found that the VAL60 fragments have on average at least one CATH structure within 0 . 6 Å of RMSD ., For all the structural descriptors we considered the VAL60 distributions are similar but not identical to the ones of real proteins , due to the fact that in our simulation we considered an homopolymer formed by only one amino acid , valine ., Taken together the data shown in Fig . 2 demonstrate our first major result: finding by molecular dynamics at an all-atom level a library of 30000 protein-like structures ., The VAL60 structures obtained in this manner , at a first sight , cannot be distinguished from folds adopted by proteins ., In order to understand how many independent structures are actually explored , and if the set contains all the known folds , a measure of the degree of similarity between two protein structures is needed ., We used the TM-align approach 23 , which gives , as three quantitative outputs , the coverage , the root mean square distance ( RMSD ) between the aligned residues , and the TM-score ( see Methods ) ., Following Ref ., 15 , we first checked if the set of structures generated by molecular dynamics reproduces all the known folds ., As a target set we here considered the CATH database 19 , that is successfully used in structural studies to classify protein folds ., Other choices were also considered ( see Text S1 ) ., For each structure in the CATH database , we searched , in the set of the 30 , 000 structures of VAL60 generated by molecular dynamics , for its most similar structure as quantified by the TM-score ., In Fig . 1-b , three CATH structures with their respective VAL60 equivalent are shown ., As shown in Fig . 3-a , for almost every CATH structure it is possible to find a VAL60 structure that is very similar ., For CATH structures of length between 55 and 65 amino acids the average coverage is 75% , and the average RMSD is of only 2 . 8 Å ., The VAL60 set reproduces , with even greater success , CATH structures of shorter length ., Instead , structures of 65 or more amino acids are reproduced less accurately , as the maximum coverage that can be attained is , by definition , smaller than their length ., However , even in these cases , the RMSD restricted to the aligned residues is small , of 3 Å or less ., Comparison of the VAL60 set with even longer chains is not considered here: the long chains can contain extra secondary structure elements that do not significantly affect the quality of the alignment but change the topological details of the fold ., The excellent capability of the VAL60 set of reproducing the known folds is confirmed by monitoring the progress of exploration as a function of the number of structures found during the simulation ., At this purpose , we assumed that a CATH structure is “found” when molecular dynamics explores a VAL60 structure whose TM-score ( with respect to the CATH structure ) is higher than 0 . 45 ., Visual inspection reveals that two structures of similar length and of relative TM-score larger than 0 . 45 are structurally and topologically similar ., In Fig . 3-b we plot , for different length classes , the fraction of CATH structures that are found as a function of the number of VAL60 structures ( which is approximately proportional to simulation time ) ., At the end of the simulation , for length L\u200a=\u200a55–65 the fraction of found structures is 86% ( 85% for L\u200a=\u200a40–55 and 78% for L\u200a=\u200a65–75 ) ., 100% of the structures of length L\u200a=\u200a40–65 are reproduced within a TM-score of 0 . 4 ., This shows that the computational setup used in this work allows us to explore the majority of the folds in nature , at least within the limited range of lengths considered ., This is the second main message of our study and confirms the results of Ref ., 15 obtained with a simpler potential energy function ., The exploration of VAL60 structures by molecular dynamics proceeds in an almost random manner , with no obvious preference for a specific class of folds or secondary structure element ., Indeed we checked that it is , on average , equally likely to find a specific CATH structure as finding a VAL60 structure for the second time ( see Methods ) ., In other words , in our sampling strategy there is no particular bias for generating a structure observed in nature ., However , one realizes that the two sets of structures , CATH and VAL60 , cannot be fully equivalent ., Indeed , according to a clustering procedure ( see Methods ) , in the simulation explores independent structures , much more than the structures in CATH ( in a length range between 40 and 75 ) ., One could argue that finding or not a one-to-one correspondence might just depend on the chosen similarity threshold 24 ., In order to quantitatively investigate this issue , we addressed the following question: Do structural descriptors exist whose distributions are different between the two sets CATH and VAL60 ?, If the answer is yes , a biased search mechanism reflecting an evolutionary pressure may be envisaged ., Otherwise a random search mechanism in a continuous structure space may be enough to account for the choice of the observed folds out of all possible structures ., While at first sight structures belonging to the VAL60 and CATH sets look indistinguishable , a more detailed analysis reveals that several VAL60 structures include a large fraction of parallel -sheets ., This secondary structure element is much less common in the CATH set restricted to ., We quantify this observation by looking at the distributions of normalized contact order ( CO ) and the contact locality ( CL ) ( see Methods ) ., The distribution of CATH is significantly restricted towards lower CO and higher CL values with respect to VAL60 ( see Fig . 4-a ) , consistent with the observation that parallel -sheets are found less frequently in CATH ., We have checked that this discrepancy is not due to the specific simulation setup ( see Methods ) ., We also checked that the CO distribution computed for the subset of VAL60 that are recognized to be similar to CATH is largely overlapping with the CO distribution for the CATH set ( see Fig . 4-b ) ., This demonstrates the consistency of the similarity measure provided by the TM-score ., We also analyzed the distribution of the CO restricted to the different structural classes ., The bias towards low CO is not effective for all- structures ( see Fig . S6 ) , whereas is active for all- and - structures ., All these results suggest that , among all possible conformations physically attainable by polypeptide chains , real protein structures were selected under a bias towards low CO ., This is the third main message of our study: As observed with the coarse grained model of ref ., 16 , there is no one-to-one correspondence between the PDB library and the ensemble of compact structures with significant secondary content ., By using atomistic simulations and a powerful enhanced sampling technique we have generated a database of structures corresponding to energy minima of a 60 amino acids polypeptide ., Clearly , the length of 60 amino acids used in the simulation does not provide a complete representation of the full protein universe , which includes a very large amount of much longer proteins ., However , our results indicate that , within the limited length range we considered , the VAL60 set is indeed representative of the space inhabited by real proteins ., In fact , this set includes all the folds existing in nature for proteins of similar size , confirming that the observed protein folds are selected based on geometry and symmetry and not on the chemistry of the aminoacid sequence 5–15 ., However , we find that the known folds form only a small fraction of the full database ., Natural folds are indistinguishable in terms of secondary content and compactness from non-natural folds , but are characterized by a relatively small contact order and a relatively high contact locality ., Why has nature made this choice ?, One can argue that , due to a higher -structure content , large CO structure could have a higher tendency to aggregate ., Another possible explanation relies on kinetic accessibility , as the contact order is known to correlate with the folding time of two-state globular proteins 25 ., Evolution might have selected the folds under the guidance of a simple principle: reducing the entanglement in the bundle formed by the protein in its folded state ., Bundles with shorter loops might be preferable , as they are explored more easily starting from a random coil ., How has nature been able to select low contact order structures ?, In order to address this issue , we investigated the role of specific amino acids in selecting a fold among the possible structures ., At this scope , we compared the correlation between potential energy and CO of the structures obtained by energy minimization of VAL60 and ALA60 ( see Methods ) ., Fig . 5 vividly demonstrates that different low energy structures may be discriminated when different sequences are mounted on all the possible “presculpted” structures 12 ., Whereas energetically VAL60 prefers structures with high CO and a large content of strands , ALA60 promotes conformations with low CO and which are rich in helices ., Evolution , possibly also guided by the kinetic bias hypothesized above , can then proceed by using a repertoire of 20 types of amino acids , to select and design the sequences which minimize the free energy of a desired structure against other competing structures ., As a final remark , we believe that the VAL60 structures and the computational procedure to generate them , also with different types of amino acids and with different lengths , may play a key role in future developments ., The availability of a rich library of possible folds and realistic decoys could allow for major advances in the two main applicative challenges in protein physics: the prediction of the native state of any given sequence and the design of the sequence folding into a desired fold ., They might be also used to check predictions in synthetic biology 26 , 27 ., Furthermore the library could be exploited to obtain models of misfolded protein structures related to neurodegenerative diseases 28 ., We have shown that generating a huge set of realistic structures is feasible with a computational analysis based only on ab-initio physico-chemical information , with no need of using knowledge-based potentials as in state-of-the-art approaches to protein structure prediction and design 29 ., Molecular dynamics ( MD ) simulations are performed using the AMBER03 20 force field and the molecular dynamics package GROMACS 30 ., Simulations are mainly performed in vacuum , but tests have been performed also in water solution ( see below ) ., The temperature is controlled by the Nose-Hoover thermostat , and the integration time step is ., In order to explore the conformational space we use bias-exchange metadynamics ( BE-META ) 18 , 31 with 6 replicas ., BE-META is a combination of replica exchange 32 and metadynamics 33 , in which multiple metadynamics simulations are performed at the same temperature ., Each replica of the system is biased with a one-dimensional metadynamics potential acting on a single collective variable ( CV ) ., The CVs are described in detail in 34 and are designed in order to evaluate by a differentiable function of the coordinates the fraction of a secondary structure element ( -helix , parallel -sheet and antiparallel -sheet ) ., For instance , for the antiparallel -sheet the variable counts how many pairs of 3-residue fragments in a given protein structure adopt the correct -conformation , measured by the RMSD from an ideal block of antiparallel formed by a pair of three residues ., We use six CVs: 3 -CVs each biasing one third of the protein , 1 anti- CV , and 2 para- CV ., The Gaussians entering in the metadynamics potential are added every ., Their height and width are and 0 . 3 ., Exchanges between the biasing potentials are allowed every ., The exchanges greatly enhance the capability of the dynamics of exploring new structures 18 , 35 ., These parameters have been optimized according to the criteria of Ref ., 36 ., The main scope of this work is exploring exhaustively the conformational space of an average length polypeptide described by a realistic potential energy function ., The final choice of simulating VAL60 in vacuum with at , and then optimizing the configurations with was taken after considering several alternatives ., We first considered performing the simulation on a 60-alanine in vacuum ( ALA60 ) , as alanine is used in Ref ., 11 ., This system was evolved using the BE-META setup described above for generating structures with a high secondary content ., However , the structures generated in this manner are too compact to be comparable with experimental structures ., Indeed , the histogram of the radius of gyration for ALA60 is peaked approximately 1 Å too low with respect to what observed for real proteins of similar length ( see Fig . S1 ) ., This is due to the relatively low steric hindrance of the side chain of ALA ., The same histogram computed for VAL60 is instead fully consistent with the distribution observed in real proteins ., We also performed test simulations of VAL60 solvated in TIP3P water at ., This system was evolved for with the same BE-META setup ., In this case structures with a high secondary content are found , but most of these structures are not independent , as the correlation time in water is much larger than in vacuum ., More importantly , the structures generated in water have on average a large radius of gyration ( see Fig . S1 ) ., This is an indication that at the system explores mainly non-compact structures ., Of course , one could perform the simulation at lower temperature , but this would lead to an even larger correlation time , making an exhaustive exploration of the configuration space too time consuming with existing computational resources ., Performing the exploration with is not strictly necessary , as test simulations performed with are also able to explore structures with a high secondary content ., However , VAL60 with has a relatively high preference for structures ( see Fig . 5 ) ., With and structures become approximately isoenergetic for VAL60 , removing a possible bias in the exploration ( see also Fig . S5 ) ., The VAL60 set was generated by molecular dynamics in vacuum at 400 K , biasing the system by metadynamics potentials aimed at producing secondary structure elements ., One wonders if the structures that are explored in this manner have protein-like topologies only because of the bias , and would fall apart in normal conditions ., In order to address this issue , for all the structures generated by molecular dynamics we performed a steepest decent ( SD ) simulation with , aimed at localizing the closest potential energy surface minimum ., For the last configuration the RMSD was calculated with respect to the initial structure ., The distribution of this quantity is shown in Fig . S2 ., Most of the structures do not drift significantly apart from the initial configuration , and the RMSD remains relatively small , within 2 Å in most cases ., Thus , we conclude that the VAL60 structures generated by molecular dynamics are close to local energy minima ., The set of structures generated in this manner form the database on which we perform the analysis ., We also checked if the structures that are generated in this manner are stable if the homopolymer chain is formed by another amino acid ., At this purpose , VAL60 structures were chosen randomly ., For each of these structure the valines were replaced by alanines ( ALA60 ) ., Following the same procedure described above , a SD simulation was run until the closest local minimum is reached ., The RMSD from the initial ALA60 configuration was calculated ., The distribution of this quantity is shown in Fig . S2 ., Quite remarkably , even if one changes the amino acid sequence from VAL60 to ALA60 the structures do not change significantly , remaining within Å of RMSD from the initial structure ., This confirms the prediction of Ref ., 11 ., The similarity between two different structures is assessed using the TM-align algorithm 23 ., This method , regardless of the primary sequence of the two proteins , attempts to align their secondary structure elements allowing insertions and deletions of residues ., The fraction of aligned residues is called coverage , and is the first measure of similarity ., Afterward , the algorithm finds the rotation and translation that minimizes the relative distance between pairs of aligned residues ( RMSD ) ., The optimal coverage and RMSD are then combined into a single similarity measure , the TM-score ., The original version of the TM-align algorithm has been modified in order to assign the secondary structure elements with more accuracy ., Instead of considering only the coordinates as in Ref ., 23 , our modified version reads for each protein the secondary structure assignment given by DSSP 37 ., When the proteins have different lengths , the length of the target protein is used in the TM-score definition 23 ., The TM-score is equal to one for two identical structures ., Two structures are considered to represent the same fold if their TM-score is greater than 0 . 45 , while for two randomly chosen structures the TM-score is approximately equal to 0 . 25 ., In order to find the independent structures we proceeded as follows: first we selected the structure with the largest number of neighbors , namely with the largest number of structures at a TM-score larger than 0 . 45 ., We assign it as the first independent structure and remove it , together with all its neighbors , from the list of structures ., We iterate this procedure until the list is empty ., In Fig . S3 we plot the number of independent structures found as a function of the number of structures explored by MD ., This data can be accurately reproduced with a double exponential fit ( ) , which allows estimating as the number of independent structures that would be explored in an infinitely long MD run ., We consider a small fraction of the MD trajectory used for generating the VAL60 dataset ., In this fraction of the trajectory independent structures are generated ., Using the rest of the trajectory , we compute the number of times that each of these structures is observed ( namely , the number of times a structure with relative TM-score larger than 0 . 45 is visited ) ., The histogram of is calculated for 20 different sets , each including 100 VAL60 structures ., Its average and standard deviation ( error bars ) are plotted in Fig . S4 ., This is compared to the same histogram computed for the CATH set with ( structures ) ., Strikingly , the two histograms are very similar , indicating that the probability of finding a CATH structure in this length range is similar to the probability of finding a VAL60 structure a second time ., Two residues are considered to be in contact when at least one pair of their heavy atoms is found at a distance smaller than 3 . 5 Å ., The contact order ( CO ) 25 is defined as the average sequence separation between contacting residues divided by the chain length ., The contact locality ( CL ) , is a structural descriptor that counts the fraction of contacting residue pairs which are formed within the same half of the chain 38 ., The total number of pairwise contacts is , where and are the contacts between residues both belonging to the half of the chain towards the N-terminus and the C-terminus , respectively , and are the contacts between residues belonging to different halves of the chain ., CL is then defined as ., One of the main results described in the work is that , on average , the VAL60 structures have higher CO than CATH structures ., In order to find out if the biasing procedure favors high CO structures we separate the VAL60 structures in two classes: low CO ( ) and large CO ( ) , and we calculate the probability to find a structure times in the simulation ( same procedure as above ) ., The two distributions with the respective error bars are shown in Fig . S5 ., From the graph , it can be concluded that the two distributions are similar but it is marginally easier for VAL60 to re-generate more times low CO structures rather than high CO ones ., Thus , the VAL60 system is able to sample low CO structures with a marginally higher efficiency ., This is possibly due to the fact that low CO structures are kinetically encountered more often in a random search guided only by a bias towards high secondary structure content ., This allows concluding that the large number of high CO structures that is obtained by molecular dynamics is not due to a bias in the sampling procedure ., The results found in Fig . 4 show that there is a bias towards low CO structures for the CATH set ., In order to find out how this bias acts for different structural classes , the CO distributions was calculated for all- structures and all- structures of CATH and VAL60 ., The results are shown in Fig . S6 ., While the bias towards low CO is present for all- structures , for all- structures it is not effective ., It is also remarkable that the CO distribution for structures in the VAL60 set that are similar to a CATH structure is very similar to the probability distribution for the all- CATH structures . | Introduction, Results, Discussion, Materials and Methods | It is currently believed that the atlas of existing protein structures is faithfully represented in the Protein Data Bank ., However , whether this atlas covers the full universe of all possible protein structures is still a highly debated issue ., By using a sophisticated numerical approach , we performed an exhaustive exploration of the conformational space of a 60 amino acid polypeptide chain described with an accurate all-atom interaction potential ., We generated a database of around 30 , 000 compact folds with at least of secondary structure corresponding to local minima of the potential energy ., This ensemble plausibly represents the universe of protein folds of similar length; indeed , all the known folds are represented in the set with good accuracy ., However , we discover that the known folds form a rather small subset , which cannot be reproduced by choosing random structures in the database ., Rather , natural and possible folds differ by the contact order , on average significantly smaller in the former ., This suggests the presence of an evolutionary bias , possibly related to kinetic accessibility , towards structures with shorter loops between contacting residues ., Beside their conceptual relevance , the new structures open a range of practical applications such as the development of accurate structure prediction strategies , the optimization of force fields , and the identification and design of novel folds . | Protein structure and biological function are determined by their sequence , but proteins of different sequence or function can share the same structure ., To rationalize this puzzling observation we explored by computer simulations the universe of all possible folds for proteins of relatively small length ., We find that nature exploits a relatively small corner of this universe ., Evolution selected this region under the guidance of a simple principle: reducing the entanglement in the bundle formed by the protein in its folded state ., This makes bundles with shorter loops preferable ., The set of structures that we make available will open a range of practical applications in biomedical sciences . | computational biology/molecular dynamics | null |
journal.pgen.1002237 | 2,011 | Genome-Wide Gene-Environment Study Identifies Glutamate Receptor Gene GRIN2A as a Parkinsons Disease Modifier Gene via Interaction with Coffee | Common disorders are thought to have both genetic and environmental components ., Genome-wide association studies ( GWAS ) have successfully identified numerous susceptibility loci for many common disorders ranging from behavioral traits such as addiction and substance abuse to infectious and immune-related disorders , age-related neurodegenerative disorders like Alzheimers , Parkinsons and macular degeneration , metabolic disorders , psychiatric disorders , and many more ( for the list and results of over 800 published GWAS see http://www . genome . gov/gwastudies ) ., Despite the success of GWAS , the heritability of common disorders cannot be fully explained by the genes that have been discovered 1 ., GWAS are built on the notion that common alleles predispose to common disorders ., Rare variants , which are probably responsible for some of the missing heritability , would not have been detected by GWAS ., Sequencing the genome and novel analytical methods will help identify the rare variants ., Another hiding place for the missing heritability is in interactions ., Genes that impact disease through interactions with other genes or environmental factors are not detected by GWAS if their main effects are small ., GWAS can only identify genes that exhibit significant main effects; genes that require the interacting factor to be included in the study to show their association with disease are missed ., Inclusion of key environmental factors in genome-wide studies is anticipated to be an important next step for deciphering the genetic structure of common multifactorial disorders ., Amassing sufficient analytic power for gene-environment studies , however , is a challenge ., Power decreases dramatically as a function of frequency of exposure , number of parameters being estimated and sample size ., Interaction studies require at least four times the sample size that standard GWAS would require to detect an effect of similar magnitude ( reviewed in 2 ) ., Yet , there are fewer datasets with both DNA and environmental exposure data than those with DNA alone , and their sample sizes are often smaller ., Parkinsons disease ( PD ) is a classic example of a common multifactorial disorder ., PD is characterized by neurodegeneration in the substantia nigra that manifests initially as a movement disorder but often leads to cognitive and psychiatric problems as well ., PD is progressive and there is no treatment currently available that could prevent or slow disease progression ., PD is the second most common neurodegenerative disease after Alzheimers disease; it affects about 5 million individuals in the 10 most populous nations and is expected to double in frequency by 2030 3 ., Until the 1990s PD was thought to be purely environmental with no genetic component ., In the last decade , numerous genes have been identified , some of which can cause PD 4 and others that are susceptibility loci 5–10 ., There are also compelling data from epidemiology that cigarette smoking and caffeinated-coffee consumption are associated with reduced risk of developing PD 11 , 12 and that exposure to environmental neurotoxins is associated with increased risk of developing PD 13 ., Thus PD is a strong candidate for studying gene-environment interactions 14 ., We conducted a genome-wide association and interaction study ( GWAIS ) using the joint test 15 for each SNPs marginal association and its interaction with coffee consumption on PD risk , followed by stratified GWAS in heavy and light coffee drinkers ( see Analytic Strategy in Materials and Methods section ) ., Our aim was to identify genes that enhance or diminish the protective effect of caffeinated-coffee for use as biomarkers for pharmacogenetic prevention and treatment ., Caffeine is an adenosine-receptor antagonist ., In animal models of PD , where administration of neurotoxins is used to destroy dopaminergic neurons mimicking PD , caffeine and selective A2A-antagonists have been shown to be neuroprotective and attenuate dopamine loss 16 ., Selective A2A-antagonists have been studied in human clinical trials and found to be safe , well tolerated and to provide symptomatic benefit for persons with PD 17 , 18; however , efficacy has not been high enough in the first generation of the drugs to meet regulatory approval for use as PD drugs ., We posit that subsets of patients with certain genotypes may respond well to a given treatment and others may not ., When they are combined the average efficacy may be insufficient for regulatory approval , while a subgroup of patients with certain genotype might still benefit substantially ., If our prediction is correct , incorporating genetics in clinical trials of PD could revolutionize PD drug development ., By examining the interaction of caffeinated-coffee with 811 , 597 SNPs in a hypothesis-free genome-wide study , we discovered GRIN2A as a novel PD modifier gene ., GRIN2A encodes a subunit of the NMDA-glutamate-receptor which is well known for regulating excitatory neurotransmission in the brain and for controlling movement and behavior ., Human Subject Committees of the participating institutions approved the study ., The Discovery dataset was nested in the NeuroGenetics Research Consortium ( NGRC ) GWAS which successfully identified known PD genes as well as a novel association with HLA 5 which has been widely replicated 10 , 19 ., For the present GWAIS , Replication samples were provided by PEG 20 ( Parkinson , Environment , and Gene ) , PAGE 21 ( Parkinsons , Genes , and Environment from the prospective NIH-AARP Diet and Health Study cohort ) , and HIHG 9 ( Hussman Institute for Human Genomics ) ., Persons with PD had been diagnosed by neurologists using standard criteria 22 , control subjects self-reported as not having PD ., Cases and controls were all unrelated , non-Hispanic Caucasian , from United States ., The NGRC cohort was clinic-based sequentially ascertained patients , PEG and PAGE were community-based incident cases , HIHG was clinic-based and self-referral cases ., The numbers of cases/controls with genotype , coffee/caffeine and key clinical and demographic data were NGRC\u200a=\u200a1458/931 , PEG\u200a=\u200a280/310 , PAGE\u200a=\u200a525/1474 , HIHG\u200a=\u200a209/133 ( Table S1 ) ., NGRC , PEG and HIHG had collected lifetime caffeinated-coffee consumption data , measured as cups per day multiplied by the number of years of consumption ( ccy ) 12 , 23 ., PAGE had daily mg caffeine intake from all caffeine-containing drinks and foods for 12 months prior to enrollment ( 1995–1996 ) and only incident PD cases diagnosed after 1997 were included in the analysis 24 ., Despite the variation in data collection , results were consistent across studies , corroborating robustness of the interaction between coffee/caffeine and GRIN2A ., We could not , and did not , attempt to distinguish the bioactive ingredient in caffeinated-coffee ., Although caffeine has been shown to be neuroprotective , there may be other ingredients in caffeinated-coffee that may affect disease pathogenesis ., To classify coffee/caffeine intake , each dataset was treated separately according to the measurements available ., The median ccy or mg was determined for controls within each dataset ( excluding those with zero intake ) and used as the cut-off for heavy drinkers ( >median ) vs . light drinkers ( 0 to ≤median ) ., The median was 67 . 5 ccy for NGRC , 74 . 0 ccy for PEG , 70 . 0 ccy for HIHG , and 237 . 8 mg/day for PAGE ., For coffee dose , quartiles were defined for each dataset using the full range from zero to maximum intake in controls ., Results shown for NGRC , PEG and HIHG are based on lifetime caffeinated-coffee consumption ., Truncating coffee use at age-at-onset or age-at-diagnosis in patients did not affect the results ., To assess the effects of caffeinated tea and soda , we performed sensitivity analysis in NGRC dataset ., Caffeinated soda and tea were commonly and equally consumed by heavy and light coffee drinkers ( soda: 80% in both heavy and light drinkers; caffeinated tea: 66% in heavy coffee drinkers and 61% in light coffee drinkers ) ., We repeated GWAIS and stratified GWAS with caffeinated soda and tea as covariates ., We also explored association of caffeinated tea and soda with PD expecting an inverse association if caffeine were the bioactive ingredient in coffee ., The source of DNA was whole blood for NGRC and HIHG , saliva for PAGE , and whole blood ( all PD and half of controls ) or saliva ( half of controls ) for PEG ., NGRC was genome-wide genotyped using Illumina HumanOmni1-Quad_v1-0_B array and achieved 99 . 92% call rate and 99 . 99% reproducibility ., GWAS genotyping and statistical quality control ( QC ) have been published 5 ., 811 , 597 SNPs ( excluding Y chromosome SNPs because they are not amenable to sex adjustment ) passed GWAS QC and were included in GWAIS ., Replication groups genotyped GRIN2A_rs4998386 ., Only one SNP was genotyped for replication; we have no other undisclosed replication results ., PEG and HIHG used ABI TaqMan assay-by-design ( C__28018721_20 ) , PAGE used Sequenom and all achieved call rates of 96%–99% ., The first step was to test the hypothesis that the effect of coffee on PD risk is affected by a gene; ie , test statistical interaction between SNPs and coffee genome-wide ., Theoretically , a test of SNP*coffee interaction would have been suitable; however , a pure test of interaction has low power; reportedly , it requires more than four times the sample size that GWAS would require to detect a main effect of similar size ( reviewed in 2 ) ., We chose the joint test of SNP main effect and its interaction with coffee as proposed by Kraft et al 15 ., We call the test GWAIS for genome-wide association and interaction study ., The main advantage of the joint test is that it does test for interaction and it has more power than pure interaction test when there is a modest SNP marginal effect ., Next we performed stratified GWAS in heavy and light drinkers to gain insight to where the interaction signal was coming from and to formulate a hypothesis for replication ., We then replicated the top signal and performed pooled analysis ., Methods for meta-analysis of the joint test are available 25 , 26; however , since we had individual level data we pooled the datasets ., The most significant result was the novel appearance , on the Manhattan plot ( Figure 1A , Figure S1 ) , of a block of linked SNPs which map to the GRIN2A gene on chromosome 16 ( Figure S2 ) ., This locus had not been detected in PD GWAS previously because its main effect is modest ., However , when considered in the context of interaction with coffee , GRIN2A surpassed all known PD-associated genes in significance including SNCA which has been the strongest association with PD in GWAS ., The signal for known PD genes were driven only by their main effects with no evidence for interaction ( Pinteraction\u200a=\u200a0 . 5–0 . 7 ) ; whereas the signals for PD-associated SNPs in GRIN2A were enhanced by SNP*coffee interaction ( Pinteraction∼10−3 ) ., The quantile-quantile ( QQ ) plot of the expected vs . observed genome-wide P values ( Figure 1B ) is also evidence for the impact of GRIN2A on PD risk ., GWAIS results described above were obtained from a test that measures the combined significance of the SNP and its interaction with coffee on risk of PD 15 ., The test has 2 df; hence when interaction is absent , GWAIS is less powerful than GWAS which has only 1 df ., Furthermore , the sample size was smaller in GWAIS because it required not only genotypes but also coffee data , which was available for 2/3 of NGRC ., Under these conditions , GWAIS produced P2df>10−6 ( Figure 1A ) for the top SNP in SNCA which had reached P\u200a=\u200a3×10−11 in NGRC GWAS 5 ., This drop in significance demonstrates the dramatic loss of power in GWAIS as compared to GWAS ., Under these conditions , GWAIS yielded P2df\u200a=\u200a1×10−6 for rs4998386 in GRIN2A ( as compared to P2df\u200a=\u200a3×10−6 for SNCA and P2df\u200a=\u200a7×10−5 for MAPT ) ., Dominant and Additive models produced nearly identical results for GRIN2A SNPs ( Table 1 ) ., Recessive model had no notable signal ( Figure S1 ) ., With one goal being pharmacogenetic applications , we were interested in genes that modulate risk in people who consume caffeine , thus we stratified the subjects as heavy drinkers or light drinkers ( light includes non-drinkers ) and performed GWAS in each group ( SNP-PD test , 1 df ) ., The sample size was now further reduced to only 512 cases and 387 controls who drank more than the median ( heavy drinkers ) and 946 cases and 544 control subjects who drank less than the median ( light drinkers ) ., As expected due to interaction , which suggests different association patterns across categories , most of the signals seen in GWAIS ( Figure 1A ) appeared within either heavy drinkers ( Figure 2 , Table 2 , Figure S1 ) or light drinkers ( Figure 3 , Table 2 , Figure S1 ) ., In heavy drinkers , the focus of this study , the most significant result was GRIN2A_rs4998386 ( P\u200a=\u200a6×10−7 ) and 11 neighboring SNPs ( P\u200a=\u200a10−5 to 10−6 , Table 2 ) ., The QQ plots for stratified GWAS also demonstrate clearly that GRIN2A is the single primary PD associated locus in heavy coffee drinkers ( Figure 2 ) : exclusion of SNCA , HLA and MAPT did not have an impact in heavy drinkers , whereas exclusion of GRIN2A nearly abolished the extreme P values of 10−5–10−6 ., No clear signals were detected in light coffee drinkers ( Figure 3 ) ., The 12 GRIN2A SNPs that were associated with PD via heavy coffee consumption had similar minor allele frequencies ( MAF\u200a=\u200a0 . 13–0 . 16 in controls ) and odds ratios ( OR\u200a=\u200a0 . 43–0 . 51 ) and were in strong LD ( Figure S3 ) ., Haplotype analysis did not strengthen the signal ., Within this gene varying CNV software tools called either no CNVs or just two CNVs in controls ., Thus , CNVs are unlikely to explain a large fraction of the phenotype variability ., We therefore selected only the SNP with the lowest P value for replication ( GRIN2A_rs4998386 ) ., Testing the association of coffee with PD in NGRC , when calculated irrespective of genotype , showed an average of 34% lower PD risk in heavy coffee drinkers than in light drinkers ( OR\u200a=\u200a0 . 66 , P\u200a=\u200a6×10−6 , Table 3 , Coffee irrespective of genotype ) ., GRIN2A , irrespective of coffee , had a modest main effect on PD in NGRC ( Table 3 , GRIN2A rs4998386 genotype irrespective of coffee ) ., A key question was if , and to what degree , GRIN2A_rs4998386 genotype modifies the effect of coffee on PD risk ( Table 3 ) : Within heavy drinkers , PD risk was 58% lower ( OR\u200a=\u200a0 . 42 , P\u200a=\u200a2×10−6 ) for rs4998386_TC , and 81% lower ( OR\u200a=\u200a0 . 19 , P\u200a=\u200a0 . 05 ) for rs4998386_TT genotype than rs4998386_CC; whereas in light drinkers genotype had no effect on risk ., Similar results were obtained for Additive and Dominant models ( Table S3 ) ., The joint effect comparing rs4998386_TC genotype and heavy coffee vs . rs4998386_CC genotype and light coffee was most dramatic , suggesting a highly significant 68% risk reduction ( OR\u200a=\u200a0 . 32 , P\u200a=\u200a7×10−11 ) in NGRC ( Table 3 , Joint effects of GRIN2A rs4998386 and coffee ) ., We used GWAIS as a means to identify genes that might enhance the inverse association of coffee with PD with the goal of carrying the discovery forward as a genetic marker for use in pharmacogenetic studies ., Hence , the replication hypothesis was specified a-priori , based on results of NGRC , as follows: “Among heavy coffee drinkers , carriers of rs4998386_T allele have lower risk of PD than carriers of rs4998386_CC genotype” ., Although this test does not reflect our most significant results , it is the test that has the clearest interpretation because it keeps the effect of coffee constant ., For example , comparing TC+heavy vs . CC+light gave larger effect size and the P value was 3-orders of magnitude lower than the specified hypothesis , however , unlike our hypothesis , the test included coffee , which would have made it difficult to draw firm conclusions about the effect of genotype on coffees inverse association with PD ., Before attempting replication , the following analyses were conducted to identify potential confounders ( Table S2 ) ., We tested the frequency of rs4998386 and coffee use across disease-specific strata and population structure ., There was no evidence for heterogeneity by presence or absence of family history of PD , age at onset , or recruitment site ., rs4998386 frequency was different between Ashkenazi-Jewish and non-Jewish individuals ( P\u200a=\u200a0 . 02 ) and across the European countries of ancestral origin ( P\u200a=\u200a3×10−3 ) in cases , but not in controls , which , PD being heterogeneous , may indicate different ethnic-specific clusters of disease subtypes as has been noted for LRRK2-associated PD 34 ., Not surprisingly , heavy coffee use was associated with smoking ( P<10−4 ) , which itself is inversely associated with PD risk independently of coffee 12 ., Adjusting for smoking , in addition to other covariates , did not change the results ( Table S4 ) ., We also repeated the analyses adjusting for caffeinated soda and caffeinated tea consumption and found the results to be robust ( Table S5 ) ., Some reports suggest persons with PD are more likely to avoid sensation seeking and addictive behaviors 35 and GRIN2A polymorphisms have been implicated in predisposition to heroin addiction 36 and smoking 37 raising the concern that our results could have been confounded if the GRIN2A SNPs identified here were associated with habitual coffee drinking ., However , there was no evidence for association between any of the GRIN2A SNPs and heavy vs . light coffee consumption in cases and controls combined ( OR\u200a=\u200a0 . 95–1 . 01 , P\u200a=\u200a0 . 61–0 . 94 ) ., See Table 3 , Table S3 ., The a-priori hypothesis for replication that among heavy drinkers GRIN2A_rs4998386_T carriers had a lower risk of PD than GRIN2A_rs4998386_CC was replicated comparing TC to CC ( excluding rare heterogeneous TT genotype ) : OR\u200a=\u200a0 . 59 , P\u200a=\u200a10−3; under Additive model ( TT vs . TC vs . CC ) : OR\u200a=\u200a0 . 77 , P\u200a=\u200a0 . 04; and Dominant model ( TT+TC vs . CC ) : OR\u200a=\u200a0 . 70 , P\u200a=\u200a0 . 01 ., Note that the Additive and Dominant models included the TT genotype which is rare and its frequency varied significantly across datasets ( Table 3 , Table S6 ) ., The TC vs . CC comparison is more robust for this reason; Additive and Dominant model are shown for completeness ., As seen in NGRC data , genotype had no effect on risk of PD among light coffee drinkers in Replication or combined data ( OR\u200a=\u200a1 . 0 , P\u200a=\u200a0 . 99 ) ., In pooled Replication ( without Discovery ) , the SNP+SNP*coffee joint test yielded P2df\u200a=\u200a2 . 3×10−3 comparing TC to CC ( excluding rare heterogeneous TT genotype ) ; P2df\u200a=\u200a0 . 12 for the Additive model , P2df\u200a=\u200a0 . 02 for the Dominant model ., The pooled analysis of Replication and Discovery with the SNP+SNP*coffee joint test yielded , P2df\u200a=\u200a1 . 9×10−7 comparing TC to CC ( excluding rare heterogeneous TT genotype ) , P2df\u200a=\u200a1 . 4×10−5 for the Additive model , and P2df\u200a=\u200a8 . 6×10−7 for the Dominant model ., In pooled data , compared to the light coffee drinkers with GRIN2A_rs4998386_CC genotype ( the group with highest risk ) , heavy coffee use ( with CC genotype ) reduced risk by 18% ( OR\u200a=\u200a0 . 82 , P\u200a=\u200a3×10−3 ) , having GRIN2A_rs4998386_T allele ( light coffee ) had no effect on risk ( OR\u200a=\u200a1 . 0 , P\u200a=\u200a0 . 99 ) , but the combination of heavy coffee use and GRIN2A_rs4998386_TC genotype was associated with a highly significant 59% risk reduction ( OR\u200a=\u200a0 . 41 , P\u200a=\u200a6×10−13 ) ( Table 3 , Joint effects of GRIN2A rs4998386 and coffee ) ., See Table 4 , Table S7 ., The array used in the study , Illumina OMNI-1 had nearly a million SNPs , which is a relatively dense coverage , but which could be further improved by imputing the SNPs that were not on the array using 1000 Genomes and HapMap data , a practice that has successfully aided many projects ., After QC , we had over 3 . 5 million imputed and genotyped SNPs per individual in NGRC , each with information score ≥0 . 95 ( measure of imputation certainty ) , and each passing standard GWAS QC ., Imputation could only be applied to NGRC ( Discovery ) because only NGRC had genome-wide data ., GWAIS and GWAS analysis of the GRIN2A region with imputed SNPs uncovered a block of densely linked SNPs embedded amongst the genotyped GRIN2A , six of which achieved P2df≤5×10−8 in GWAIS ( Table 4 ) ., The interaction term was OR\u200a=\u200a0 . 44 , P\u200a=\u200a4×10−5 ( Table 4 ) ., In GWAS conducted in heavy coffee drinkers , 12 SNPs achieved P\u200a=\u200a3×10−8 to 5×10−8 with OR\u200a=\u200a0 . 41–0 . 42 ( Table 4 ) ., In a genome-wide gene-environment study we identified GRIN2A as a genetic modifier of the inverse association of coffee with the risk of developing PD ., The discovery was made in NGRC , and replicated in independent data ., Risk reduction by heavy coffee use , which was estimated to be 27% on average , was genotype-specific and varied according to GRIN2A genotype from 18% ( P\u200a=\u200a3×10−3 ) for individuals with rs4998386_CC genotype to 59% ( P\u200a=\u200a6×10−13 ) for those with rs4998386_TC genotype ., When coffee intake was categorized in four doses , the dose trend was more prominent in individuals with rs4998386_T allele than those with rs4998386_CC genotype , with the 3rd and 4th quartiles exhibiting only 11% and 39% risk reduction for rs4998386_CC carriers , vs . 37% and 66% for rs4998386_T carriers ., With imputation we uncovered a block of GRIN2A SNPs not included on the genotyping array , which achieved P\u200a=\u200a3×10−8 to 5×10−8 ., We propose GRIN2A as a new modifier gene for PD , and posit that if coffee-consumption is considered , GRIN2A may prove to be one of the most important PD-associated genes to have emerged from genome-wide studies ., We base this suggestion on statistics , biology and the potential for immediate translation to clinical medicine , as we discuss below ., GRIN2A had not previously been tested as a candidate gene for PD , and was not detected in PD GWAS which have all been examining gene main effects without considering interactions with relevant environmental exposures ., The most significant and consistently replicated main effects detected to date are for SNCA , MAPT and HLA ., Here we added , for the first time , a common and relevant environmental exposure ( coffee ) to a genome-wide study ., Inclusion of coffee allowed GRIN2A to rise to the top ., In the gene-environment ( GWAIS ) model , GRIN2A surpassed SNCA , MAPT and HLA in statistical significance ., Among heavy coffee drinkers , the impact of GRIN2A on PD risk ( measured as OR ) was 50% greater , and 2 to 5 orders of magnitude more significant ( measured as P value ) than the strongest associations reported for SNCA , MAPT or HLA ., This study is proof of concept that inclusion of environmental factors can help identify disease-associated genes that are missed in SNP-only GWAS ., GRIN2A is an important gene for the central nervous system ., Accelerated evolution of GRIN2A in primates is said to have contributed to the dramatic increase in the size and complexity of the human brain which defines human evolution 38 ., GRIN2A encodes a subunit of the N-methyl-D-aspartate-2A ( NMDA ) glutamate receptor ., It is central to excitatory neurotransmission and the control of movement and behavior 39–41 ., The literature suggest imbalances in NMDA-dependent neurotransmission contribute to neurodegeneration in PD , possibly through massive influx of calcium and impaired mitochondrial function leading to apoptosis; and/or disruption of glutamate-mediated autophagy which is implicated in degradation and removal of proteins like α-synuclein ( see 42 for review ) ., The portion of intron 3 containing SNPs with the most significant associations ( from base pair 9978046 to base pair 10128367 , Table 1 , Table 2 , and Table 4 ) includes numerous transcription factor binding sites and two peaks of enhanced histone H3K4 mono-methylation ( http://genome . ucsc . edu ) 43 ., Polymorphisms throughout this region could therefore disrupt regulatory elements , potentially leading to variation in levels of GRIN2A transcript ., GRIN2A is expressed at high levels in the brain , most notably in the subthalamic nucleus ( STN ) 44 ., Pharmacologic inhibition of STN with an NMDA-antagonist reduces nigral neuron loss in a rodent model of PD 45 ., Deep-brain-stimulation , which also targets STN , is an effective surgical symptomatic therapy for PD ., The other piece of this finding is coffee/caffeine ., Our study was not designed to distinguish the active ingredients in coffee ., However , we note that the largest replication study ( PAGE ) measured specifically the caffeine intake in mg from all food sources ( drink , food , and chocolate ) and replicated our hypothesis and interaction robustly ., We also found trends for inverse association of tea and soda with PD , and interestingly , the varied effect size and strength of association was consistent with the relative amount of caffeine in each drink ( Table S5 ) ., Thus , our data are consistent with experimental observations that caffeine is neuroprotective ., Caffeine is an adenosine A2A-receptor antagonist ., A2A-receptor enhances calcium influx via NMDA 41 and A2A-receptor antagonists are neuroprotective in animal models of PD; they attenuate excitotoxicity by reducing extracellular glutamate levels in the striatum 46 , 47 ., Thus interaction between coffee/caffeine and GRIN2A is biologically plausible , and can help formulate testable hypotheses towards a better understanding of the disease pathogenesis ., GRIN2A genotyping may be useful for pharmacogenetic studies ., Genetics has not yet entered drug development for PD but the time is here ., We now have several susceptibility loci ( SNCA , MAPT , HLA , BST1 , PARK16 , GAK 5–10 ) that can help identify individuals who are at moderately increased risk of developing PD ., We also have at least one neuroprotective compound ( coffee/caffeine ) which can be pharmacologically modified to alleviate its undesirable side effects ., GRIN2A genotyping might also inform treatments for people who already have PD ., L-DOPA , the primary PD drug for 40 years , does not slow disease progression and has serious side effects ., Clinical trials for new PD drugs have not found drugs that surpass the symptomatic benefits of L-DOPA ., There have been numerous drug trials for glutamate-receptor blockers as well as for selective A2A-receptor antagonists ., Most were shown to be safe , well tolerated and beneficial 17 , 18 , 48; however , the majority did not reach the regulatory threshold for efficacy to be approved as PD drugs ., We wonder if some of these clinical trials will succeed if patients are subdivided by GRIN2A genotype ., We acknowledge the distinction that the present study examined risk of developing PD; whereas clinical trials have thus far aimed for symptomatic improvements in patients ., Nonetheless , there are sufficient parallels to suggest that GRIN2A genotype might also influence efficacy of glutamate-receptor antagonists and A2A-receptor antagonists ., This is a simple and inexpensive hypothesis that can be tested in future , ongoing and even closed clinical trials that have banked DNA ., Common non-coding variants in GRIN2A have been associated with Huntington disease ( HD ) 49 , 50 and schizophrenia 51 , and rare mutations have been described in patients with neurodevelopmental phenotypes 52 ., Schizophrenia is associated with a ( GT ) n repeat in the GRIN2A promoter that may increase disease risk by suppressing gene expression 51 ., Three GRIN2A SNPs have been associated with onset-age of HD; they are conserved and reportedly tag a binding site for CCAAT/enhancer-binding protein 49 , 50 ., HD and PD are both neurodegenerative movement disorders , thus the possibility of a common genetic element was of interest ., The reported HD-associated GRIN2A SNPs , rs1969060 , rs8057394 and rs2650427 , were not on the genotyping array but were imputed with high fidelity ( information score >0 . 99 ) ., They map within the 150 kb region identified here for PD , they are in strong LD with PD-associated SNPs defined by D ( 0 . 48–1 . 0 ) but not by r2 ( 0–0 . 33 ) ( Figure S4 ) ., We tested the HD SNPs for association with onset age and risk of PD in NGRC while conditioning on the neighboring top PD SNP ( rs4998386 ) ., One HD SNP , rs8057394 , yielded OR\u200a=\u200a0 . 85 , P\u200a=\u200a0 . 02 for PD overall; OR\u200a=\u200a0 . 79 , P\u200a=\u200a0 . 04 for heavy coffee drinkers; and OR\u200a=\u200a0 . 90 , P\u200a=\u200a0 . 24 for light coffee drinkers ., We found no other evidence for association of HD SNPs with PD , including when we jointly tested HD SNPs and possible interaction with coffee SNP+SNP*coffee on risk or onset of PD ., Conversely , we retested , in NGRC , the association of top genotyped PD SNP ( rs4998386 ) with PD , conditioning on HD SNP ( rs8057394 ) and found it to be robust ( P2df\u200a=\u200a8×10−6 ) ., Unlike GWAS , which is now a fully standardized practice , there is no established protocol for testing gene*environment interaction on a whole-genome scale ., Our strategy of starting with the joint test ( GWAIS ) and following up with GWAS in subgroups stratified by exposure was driven by the aims of our study ., In Table S7 we present a side-by-side comparison of the results for the top GRIN2A SNPs ( P<10−5 ) , when analyzed for main effect ( GWAS ) , for interaction , with Krafts joint test , and in GWAS stratified by exposure ., Amassing a large enough sample size for GWAIS is challenging ., GWAIS requires larger sample sizes than GWAS yet there exist fewer samples that have data on relevant environmental exposures in addition to DNA and phenotype ., To our knowledge , NGRC is the largest genetic study of PD that has collected exposure data ., No other publically available PD GWAS has coffee data , eliminating the possibility of in-silico replication ., We were able to identify and get access to only 3 datasets that had DNA and coffee , giving us a total sample size of 393 cases and 905 controls to replicate the GRIN2A effect in heavy coffee drinkers ., In contrast , replications and meta-analyses for gene-only GWAS now have over 17 , 000 PD cases and controls 10 ., We detected the known and confirmed PD-associated genes ( SNCA , MAPT and HLA ) in GWAIS but at much lower significance levels than in GWAS because of the smaller sample size with coffee data and the added degree of freedom in GWAIS ., It is noteworthy , however , that at P2df\u200a=\u200a10−6 , GRIN2A surpassed all known PD loci in significance ., With the aid of imputation , we achieved P\u200a=\u200a3×10−8 for a 2 . 4-fold difference in genotype specific effect of coffee on risk of PD ., Importantly , we were able to replicate the hypothesis that we set out a-priori based on discovery . | Introduction, Materials and Methods, Results, Discussion | Our aim was to identify genes that influence the inverse association of coffee with the risk of developing Parkinsons disease ( PD ) ., We used genome-wide genotype data and lifetime caffeinated-coffee-consumption data on 1 , 458 persons with PD and 931 without PD from the NeuroGenetics Research Consortium ( NGRC ) , and we performed a genome-wide association and interaction study ( GWAIS ) , testing each SNPs main-effect plus its interaction with coffee , adjusting for sex , age , and two principal components ., We then stratified subjects as heavy or light coffee-drinkers and performed genome-wide association study ( GWAS ) in each group ., We replicated the most significant SNP ., Finally , we imputed the NGRC dataset , increasing genomic coverage to examine the region of interest in detail ., The primary analyses ( GWAIS , GWAS , Replication ) were performed using genotyped data ., In GWAIS , the most significant signal came from rs4998386 and the neighboring SNPs in GRIN2A ., GRIN2A encodes an NMDA-glutamate-receptor subunit and regulates excitatory neurotransmission in the brain ., Achieving P2df\u200a=\u200a10−6 , GRIN2A surpassed all known PD susceptibility genes in significance in the GWAIS ., In stratified GWAS , the GRIN2A signal was present in heavy coffee-drinkers ( OR\u200a=\u200a0 . 43; P\u200a=\u200a6×10−7 ) but not in light coffee-drinkers ., The a priori Replication hypothesis that “Among heavy coffee-drinkers , rs4998386_T carriers have lower PD risk than rs4998386_CC carriers” was confirmed: ORReplication\u200a=\u200a0 . 59 , PReplication\u200a=\u200a10−3; ORPooled\u200a=\u200a0 . 51 , PPooled\u200a=\u200a7×10−8 ., Compared to light coffee-drinkers with rs4998386_CC genotype , heavy coffee-drinkers with rs4998386_CC genotype had 18% lower risk ( P\u200a=\u200a3×10−3 ) , whereas heavy coffee-drinkers with rs4998386_TC genotype had 59% lower risk ( P\u200a=\u200a6×10−13 ) ., Imputation revealed a block of SNPs that achieved P2df<5×10−8 in GWAIS , and OR\u200a=\u200a0 . 41 , P\u200a=\u200a3×10−8 in heavy coffee-drinkers ., This study is proof of concept that inclusion of environmental factors can help identify genes that are missed in GWAS ., Both adenosine antagonists ( caffeine-like ) and glutamate antagonists ( GRIN2A-related ) are being tested in clinical trials for treatment of PD ., GRIN2A may be a useful pharmacogenetic marker for subdividing individuals in clinical trials to determine which medications might work best for which patients . | Parkinsons disease ( PD ) , like most common disorders , involves interactions between genetic make-up and environmental exposures that are unique to each individual ., Caffeinated-coffee consumption may protect some people from developing PD , although not all benefit equally ., In a genome-wide search , we discovered that variations in the glutamate-receptor gene GRIN2A modulate the risk of developing PD in heavy coffee drinkers ., The study was hypothesis-free , that is , we cast a net across the entire genome allowing statistical significance to point us to a genetic variant , regardless of whether it fell in a genomic desert or an important gene ., Fortuitously , the most significant finding was in a well-known gene , GRIN2A , which regulates brain signals that control movement and behavior ., Our finding is important for three reasons: First , it is a proof of concept that studying genes and environment on the whole-genome scale is feasible , and this approach can identify important genes that are missed when environmental exposures are ignored ., Second , the knowledge of interaction between GRIN2A , which is involved in neurotransmission in the brain , and caffeine , which is an adenosine-A2A-receptor antagonist , will stimulate new research towards understanding the cause and progression of PD ., Third , the results may lead to personalized prevention of and treatment for PD . | medicine, biology | null |
journal.pcbi.1007229 | 2,019 | Predictions of time to HIV viral rebound following ART suspension that incorporate personal biomarkers | Antiretroviral therapy ( ART ) for HIV infection can very effectively control the infection and hold the amount of circulating virus below the level detectable by clinical assays , improving both the quality and length of life ., ART suspension generally is followed by HIV rebound to high viral loads 1 , and consequently the standard of care for people living with HIV ( PLWH ) is to maintain life-long ART ., However , there is significant heterogeneity in rebound times ., In a pooled analysis of participants from six AIDS Clinical Trials Group ( ACTG ) analytic treatment interruption ( ATI ) studies to identify predictors of viral rebound , Li et al . reported widely varying times to viral rebound , with a significant number of participants maintaining viral suppression to undetectable levels for up to 2 or more months in the absence of ART 2 ., In a follow-up study , Li and his team identified a cohort of post-treatment controllers ( PTCs ) from these ATI studies , who maintained viral loads ≤ 400 HIV RNA copies/mL for ≥24 weeks 3 , 4 ., Previous reports of these rare PTCs include the VISCONTI cohort , 14 PLWH who initiated ART within three months of their estimated date of infection who were able to control HIV infection for a prolonged period after stopping ART 5 ., Results from the VISCONTI study and others suggest that PTCs may control HIV by a mechanism distinct from that of spontaneous HIV controllers 6 , 7 ., However , the factors that mediate delayed timing of HIV rebound are not well understood ., Since ART comes with a number of drawbacks including side-effects and cost , the search for biological indicators ( biomarkers ) of lasting ART-free HIV remission has become a priority in HIV cure research 8 , 9 ., Studies have already begun to bear fruit , with recent studies revealing a variety of immunological biomarkers for delayed rebound and infection control 2 , 3 , 10–13 ., While such studies are informative , they offer limited insight into the mechanisms underlying viral rebound or post-treatment control ., Mechanistic modeling inference has an established history of advancing our understanding of HIV 14 , 15 ., In this study we combine data on markers associated with rebound identified by Li et al . 2 with mechanistic mathematical models to gain deeper insight into mechanisms of viral rebound ., Modeling within-host HIV infection and treatment is a well-established field 14 , 16–24 ., By fitting models to clinical data , many parameters describing HIV dynamics such as the viral clearance rate , the infected cell death rate , and the viral burst size have been estimated 25–27 ., Existing models have mainly focused on the kinetics of early infection and the effects of treatment ., A few papers have focused on HIV control and the time to viral rebound after treatment cessation ., These include those of Hill et al . 28 , 29 , Pinkevych et al . 30 , 31 and Fennessey et al . 32 , in which the authors all assume viral rebound is the outcome of latent cell activation ., Pinkevych et al . used data from treatment interruption trials to provide the first estimates of latent cell activation rates that lead to observable viremia 30 , 31; using a related approach Fennessey et al . investigated SIV viral rebound in macaques infected with barcoded virus , to generate more detailed insights into viral rebound 32 ., However , this modeling does not account for individual-level heterogeneity in viral rebound dynamics 33 ., Hill et al . used continuous time branching processes , which are well-suited for small populations , to model within-host viral rebound dynamics ., The primary results in Hill et al . 28 , 29 are estimates of viral rebound time distributions , used in combination with careful and thoughtful consideration of within-host parameters to evaluate the needed efficacy of therapeutic agents that one day may be able to reduce the latent reservoir ., In their model , Hill et al . assumed that latently infected cells may die or activate and that newly activated cells can die or generate infected cell offspring that are infected , as a proxy for tracking virus that in turn infects new cells 28 , 29 ., Thus , Hill et al . assumed that average viral growth immediately following viral recrudescence is , on average , exponential , which may not be the case ., In this study , we take up the hypothesis that latent cell activation causes viral rebound in the short term ( <60 days ) 28 , 29 , but that significantly delayed rebounds are associated with additional mechanisms of infection control , such as anti-HIV immune responses 23 ., For example , there is evidence that T-cell exhaustion markers are predictive of shorter time to viral rebound 13 and that levels of HIV-specific T cell responses is associated with viral load after ATI 11 ., We focus on short-term delays ., We fit a simple stochastic model of viral rebound extended from our previous studies 21 , 24 to the viral rebound data from the ACTG ATI studies 2 ., In contrast to Pinkevych et al . 30 we address uncertainty in latent reservoir rebound dynamics , by modeling the time between a “successful” latent cell activation and detectable viremia stochastically and given by one of a variety of probability density functions ., We integrate into our model biomarkers with observed impact on time to viral rebound , e . g . , an individual’s expressed HIV reservoir , i . e . , levels of cell-associated HIV RNA ( HIV CA-RNA or CA-RNA ) , and ART regimen pre-analytic treatment interruption ( ATI ) 2 , 34 ., We discuss biological insight offered by parameter estimates from data , in particular on the average rate of latent cell activations that cause viral rebound 30 , 31 , 33 ., Our model output is a cumulative probability density function for the probability of an individual’s viral rebound at time t ., The output can be used for ATI clinical trial design; in particular , one can derive from our modeling a viral load testing schedule for participants to meet study objectives ., Written informed consent was provided by all study participants for use of stored samples in HIV-related research ., This study was approved by the Pennsylvania State University Institutional Review Board , the Los Alamos National Laboratory Institutional Review Board , and the Partners Institutional Review Board ., The description of the data we employ and associated collection methodologies are fully explained in 2 ., Briefly , participants in six ACTG ATI studies ( ACTG 371 35 , A5024 36 , A5068 37 , A5170 38 , A5187 39 , and A5197 40 ) were included if they were on suppressive ART , received no immunologic interventions ( e . g . , therapeutic vaccination , interleukin-2 ) , and had HIV-1 RNA less than 50 copies/ml at the time of ATI ( N = 235 participants ) ., We restricted the data we analyzed to the participants who showed viral rebound ≤ 60 days after ART cessation , in part to accommodate model simplifications , such as our assumption that the latent reservoir size remained constant from ATI to the time of viral rebound ( see Model , below ) ., Of the N = 210 participants who rebounded within 60 days of ATI , N = 84 met the following additional criteria for further study:, 1 ) had peripheral blood mononuclear cells ( PBMCs ) and plasma available for HIV reservoir quantification while on ART prior to the ATI and, 2 ) had cell-associated HIV-1 RNA ( HIV CA-RNA ) above the level of detection at ATI ., Cell-associated DNA was also measured but did not have a significant association with time to viral rebound 2 and is neglected in our analysis ., Finally , Li et al . noted that viral rebound delays were greater in study participants whose pre-ATI ART regimen contained non-nucleoside reverse transcriptase inhibitors ( NNRTIs ) 2 ., We therefore distinguish between regimens containing a NNRTI ( 50/84 study participants ) and those that do not ( 34/84 study participants ) ., Early after treatment interruption , most studies reported weekly viral load measurements , with the exception of A5170 ., In total , in the subset of study participants we study here ( N = 84/235 ) , 41 participants had approximately weekly or more frequent viral load measurements , while the remaining participants had viral loads measured more frequently than monthly , with a median of 7 days ( range 1-35 days; interquartile range ( IQR ) 6-24 days ) ., The timing of viral rebound was defined as sustained viral loads of at least 200 HIV RNA copies/mL ., Viral load data is shown in Fig 1a ., In this present study we model the time of viral rebound , which occurs at some point between a study participant’s last undetectable and first detectable viral load measurement ( threshold of detection 200 HIV RNA copies/mL ) ., We will therefore use those time points to estimate model parameters for each ATI study participant ., The times of last undetectable measurement and first detectable measurement are shown in Fig 1b as line segments spanning the detection window per study participant , with color indicating whether the ART regimen included ( red ) or excluded ( blue ) NNRTIs ., Although the median time between viral load tests up to the time of detectable viremia , across the 84 study participants , is 7 days , the median time window between the last undetectable measurement and the first detectable viral load measurement , shown in Fig 1b , is 20 days ( range 4-35 days; IQR 7-27 days ) ., We assume that activation of latently infected cells drives viral rebound 29 , 30 and model the ensuing dynamics as illustrated in Fig 2 ., Not all latently infected cell activations cause viral rebound ., We assume that activation is followed by rounds of viral replication , which may cause viral populations to grow to detectable levels , thereby causing viral rebound , but may also die out ., We define q as the probability of extinction , i . e . , the probability that the rounds of viral replication following latent cell activation die out , that is , that the activation of a latently infected cell does not cause viral rebound ., We further define a “successful latent cell activation” as one that does cause viral rebound ., In the time preceding a successful latent cell activation , we envision viral dynamics similar to those modeled in 24 , with potentially many latent cell activations followed by a few rounds of viral replication , with the lineages ultimately going extinct ., We also assume that any activations pre-ATI and resultant lineages go extinct as drug is still restricting viral spread ( see Incorporating rebound indicators and and Discussion ) ., Finally , we assume that there is a delay between successful latent cell activation and detectable infection , where “detectable infection” corresponds to study participant viral loads exceeding 200 HIV RNA copies/mL ., There is some debate as to the dynamics following latent cell activation ., For example , in vitro observations suggests that an activated latently infected cell may produce significantly less virus than a productively infected cell 41 ., Further , following the application of latency reversing agents , latently infected cells dynamics may not conform to the observed dynamics of productively infected cells 42 , and latently infected cells may divide before they get fully activated and produce virus 28 , 41 ., We therefore avoid the common assumption that an activated latently infected cell is the same as a productively infected cell 21 , 24 , 28 ., The need of target cells to infect in the proximity of an activated latently infected cell may also pose a challenge in initiating detectable infection in vivo ., Finally , heterogeneity in viral growth rates ( once there is enough virus for exponential growth ) among individuals , due to difference in the infecting virus and host restriction factors , is also a factor in the early dynamics 30 , 31 , 33 ., Because the dynamics of latent cell activation and infection spread before a detectable level of viremia is attained are unknown , we absorb these dynamics into a delay-time distribution , D ( t ) ( Fig, 2 ) reflecting these various sources of heterogeneity ., We will test differing assumptions on this delay , for example taking a fixed or distributed delay ( e . g . a lognormal distribution ) ., We restrict ourselves for now to short-term viral rebound ( ≤ 60 days ) ., Over that short time period , we can assume that the latent reservoir size is approximately constant with value L0 ( <3% estimated reduction , assuming that while the virus is undetectable the latent reservoir continues to decay with a half-life of 44 months 43 , 44 ) ., We assume that latently infected cells are activated at an average rate a , so activated cell influx occurs at the constant rate aL0 ., In general , we anticipate variability in the activation rate a; for example , most latently infected cells are memory cells 45 , and activation may depend on encounters with cognate antigen , whose rates may be expected to vary according to the rarity of the associated pathogen ., However , per the law of large numbers , given the large latent reservoir size in most individuals 46 , the time to detection will approximately depend on the average activation rate ., Mathematically we employ a multi-type branching process framework to derive an expression or viral rebound at time t ., We use this probability to ultimately derive likelihood functions and fit the model to data ., To sum up , in our model we assume that heterogeneity in observations of viral rebound across individuals result from four components ., Two depend on the individual study participant and derive from observed correlates of time to viral rebound:, ( i ) The replication-competent reservoir size L0 , which we will assume is reflected in the HIV CA-RNA level 2 , 34 , and, ( ii ) the probability q that the activation of a latently infected cell does not cause viral rebound , which may be affected by the pre-ATI ART regimen 2 ., The remaining two components arise from stochastic within-host dynamics:, ( iii ) the rate of latent cell activations that are “successful” , which we model as a Poisson rate , and will result in an exponential distribution in time to first successful activation , and, ( iv ) the stochastic delay between successful activation and detection , which we model using different stylized distributions ., We use the Davidon-Fletcher-Powell optimization algorithm to estimate model parameters as ( 1 − q0 ) , k , and parameters associated with the delay distribution , for our viral rebound model that neglects NNRTI status , Eq ( 1 ) , and accounts for NNRTI status , Eq ( 2 ) ., A summary of these parameter estimates are provided in Tables 2 and 3 , respectively , with complete details provided in S1 and S2 Tables , respectively ., We use the Akaike Information Criterion ( AIC ) to compare how well the models explain the data ., We show in Fig 7a predictions on the mean viral rebound time as a histogram across ATI study participants , depending on HIV CA-RNA and sorted by NNRTI status ., Our model predicts that the mean time to viral rebound is delayed in individuals including NNRTIs in their pre-ATI ART regimen ., This is not a surprise , as we were motivated to include the effects of NNRTIs by the observations of statistically significant delay in 2 ., However , our model predictions offers additional nuance: the variation in time to rebound is also larger ., Fig 7b shows a histogram of model-predicted standard deviations in time to viral rebound across the study population , again sorted by NNRTI status ., The increased variability may be explained by different NNRTI drugs and individualized differences in rates of drug metabolism ., We recover similar results using parameter estimates derived from other delay distribution assumptions ( Table 1; not shown ) ., The wider variation suggests that rebound times in PLWH taking NNRTIs are less predictable , and in the context of clinical trials , the inclusion of individuals on an NNRTI-based regimen may alter sample size and power calculations ., We can use our parameter estimates to predict a distribution in average time to successful activation across the 84 individuals ., For clarity we stress that “average” is at the individual level , since the latent reservoir is a heterogeneous population of cells ., The reservoir is primarily composed of latently infected memory cells , of which there are several types , e . g . , central memory , transitional memory and effector memory 44 ., Investigations of latent reservoir decay after initiation of therapy show multiple decay phases 57–59 , which suggest a heterogeneous population of latently infected cells with different half-lives ., Further , a memory cell is only activated when it encounters its cognate antigen , e . g . a bacterial or viral peptide that it recognizes ., Finally , recent evidence suggests that the reservoir is in part made up of clonal populations 60 ., Therefore the activation time will vary across latently infected memory cells , depending on the rate at which an individual’s immune system is challenged with different antigens ., However , if the number of cells is large , as we expect it to be in PLWH , rebound times are well described by the average ., Our model predicts that the average frequency of successful activations in an individual is given by as ( 1 − q ) log10 ( CA-RNA ) ., Fig 8 shows a histogram for the predicted time to successful activations , 1/as ( 1 − q ) log10 ( CA-RNA ) , across the ATI study population , assuming a Weibull-distributed ( Fig 8a ) or fixed ( Fig 8b ) detection delay ., Table 4 gives our model-predicted frequency of successful reactivation from latency , depending on the delay distribution assumption , with model-predicted means and 5th , 50th ( median ) , and 95th percentiles ., Previous modeling by Pinkevych et al . 30 estimated that the average frequency of successful reactivation from latency is about once every 6 days , and a range of 5-8 days ., The result of Pinkevych et al . 30 addressed neither potential heterogeneity in the unclear latent cell kinetics post-activation 41 , 42 , 61 , nor heterogeneity in viral growth rates as a source of rebound delays 31 , 33 ., In our modeling , when we neglect heterogeneity by taking a fixed delay between successful activation and infection detection , we predict an average of about 5 days ( 90% confidence interval 3-8 days ) , on par with the results of Pinkevych et al . However , when we account for that heterogeneity via alternate , i . e . , non-fixed delay densities D ( t ) , we recover a shorter average frequency of successful reactivation from latency , with successful activations occurring on average every 2-4 days ( see Table 4 ) ., Our result also adds nuance to previous estimates ., While we estimate that the mean frequency of successful activation from latency is once every 2 ( Weibull distributed detection delay ) to 5 ( fixed detection delay ) days , we also report the population range of estimated frequencies within the 5th and 95th quantiles at most once every 3-8 days , approximately , neglecting heterogeneity and assuming a fixed detection delay , and at least 1-3 days , approximately , assuming a Weibull-distributed delay ., For the purposes of further calculation , we can also use these data to estimate a population-level distribution for an individual’s average frequency of successful reactivation from which we can sample , see S2 Fig . We find that the data shown in Fig 8 is best described by a lognormal distribution , tact ∼ Lognormal ( μ , σ2 ) with μ = 0 . 74 ( standard error 0 . 03 ) and σ = 0 . 29 ( standard error 0 . 02 ) with a Weibull-distributed delay , which gives a mean average frequency of successful reactivation of once every 2 . 2 days with 90% confidence interval ( 1 . 3 , 3 . 4 ) days ( S2a Fig ) , and μ = 1 . 61 ( standard error 0 . 03 ) and σ = 0 . 30 ( standard error 0 . 02 ) , neglecting heterogeneity and assuming a fixed delay , which gives a mean of once every 5 . 2 with 90% confidence interval ( 3 . 1 , 8 . 1 ) days ( S2b Fig ) ., We envision the primary use of our parameter estimation to be ATI clinical trial design ., Our model predicts a probability density function for a study participant’s time to viral rebound following ATI , depending on that participant’s pre-ATI log10 ( HIV CA-RNA ) with level and ART regimen ., We can use the predictions to plan testing intervals to capture rebound times to within study-objective specificity ., We use tools from survival analysis , treating 1-PVR ( t ) , 1- ( cumulative probability of viral rebound function ) , as the survival function , S ( t ) = 1 − PVR ( t ) ., We have developed a simple stochastic model to predict the time to viral rebound for people living with HIV ( PLWH ) who undergo analytic treatment interruption ( ATI ) ., Our model predictions take the form of probability distributions in time , which can be interpreted as survival functions ., Our model integrates PLWH-specific data to individualize predictions based on ( 1 ) pre-ATI ART regimen and ( 2 ) pre-ATI HIV CA-RNA , both shown to be associated with times to viral rebound 2 , 3 , 34 ., Thus it distinguishes itself from previous studies , which have emphasized population average predictions 28–30 ., In our modeling we focused on short-term viral rebound following ATI , which we restrict to ≤ 60 days ., We used our model , parametrized with ATI study participant data 2 , to provide a population distribution of “successful” latent cell activation rates , i . e . , the rate of latent cell activations that induce viral rebound ., We recover an average frequency of activations leading to viral rebound of approximately 2-4 days , depending on our assumption on the delay between activation and the increase of viremia to detectable levels ., The most appropriate delay distribution cannot be resolved with existing data ., Our estimate of the successful activation rate is shorter than that of Pinkevych et al . 30 who estimate an average of about 6 days between successful activations , and a range of 5-8 days 30 ., Pinkevych et al . modeled viral rebound by assuming that the number of study participants controlling HIV at time t after ART cessation is exponentially decaying in time , and attribute that decay rate to latent cell activation ., With data from treatment interruption trials , they provided the first estimates of latent cell activation rates that lead to observable viremia 30 ., While our parameter estimation approach differs , the primary reason we obtain a shorter frequency of latent cell activation is that Pinkevych et al . did not explicitly address heterogeneity in the events leading to viral rebound ., This heterogeneity comes from many sources including the kinetics of events that occur within a latent cell post-activation 41 , 42 , 61 , including bursty transcription from the HIV-1 promoter that can lead to toggling between latent and pre-productive infection 64–66 and heterogeneity in subsequent viral growth rates 31 , 33 ., We account for heterogeneity from all sources , via the delay distribution D ( τ ) , and thus we refine their previous estimate ., However , our activation dynamics and delay distribution ignore any host factors , such as HLA allelles , that may generate inter-individual variability in time to detectable viremia ., These and other host factors may become increasingly important as we strive to improve personalized predictions of viral rebound distributions ., We hypothesize that viral rebounds occurring after many months , or even years 2 , 67 , 68—or not at all 5 , i . e . , post-treatment control—are associated with host mechanisms , such as immune responses 2 , 5 , 13 , 23 , 69 ., Markers of T-cell exhaustion are associated with times to viral rebound 13 ., Li et al . also noted that study participants treated early ( within 6 months of exposure to HIV ) showed later post-ATI viral rebound than those treated during the acute phase of infection 2 ., The recent observation of rebound following ATI delayed by 7 . 4 months , in an individual treated within an estimated 10 days of exposure to HIV 70 , is consistent with previous observations that early ART treatment is associated with delayed viral rebound timing and increased chances of post-treatment control 2 , 3 ., Delayed ART initiation appears to decrease the chances of sustained post-treatment viral control , potentially due to the expanding diversity of the HIV reservoir 71 , immune exhaustion 72 , or increasing CTL escape mutations that will diminish the effectiveness of cell-mediated immune responses 73 ., As a consequence of our hypothesis , predictions of late viral rebounds may require more sophisticated models ., In this present study , modeling viral rebound as a consequence of viral replication engendered by latent cell activation , we excluded data from ATI study participants whose viremia returned only after many months or years 2 ., The current model acts as a necessary foundation upon which immunologic data can be incorporated when they are available to model post-treatment control ., However , late viral rebounds form only a minority of dynamics following ATI , just 25 or ∼10% of 235 ATI study participants in 2 , and thus our modeling , which predicts the probability of viral rebound at time t for PLWH following ATI in the short term , describes post-ATI dynamics in the majority of individuals ., We can therefore reasonably use our modeling to aid in ATI clinical trial design , in particular determining post-ATI testing frequency , according to study objectives ., If using our model for study design and implementation , we would advise reevaluation of the few individuals who achieve 60 days with no rebound with a more sophisticated approach including testing for immunological markers of HIV control 2 , 5 , 13 , 23 , 69 ., From our modeling we can also identify gaps in data that may be invaluable in improving modeling insights into viral rebound and control ., In creating our individualized predictions , we focused on some of the first biomarkers identified to be associated with delays in viral rebound 2; work ongoing in identifying further covariates of control 2 , 3 , 10–13 will aid in further refining models of HIV rebound or control 28–30 , 32 , 74 ., When commenting on clinical trial design we must acknowledge the practical limitations , since these studies rely on PLWH who take time out of their own lives to regularly get tested ., Our data shows a median of 12 clinic visits per patient , with many making upwards of 30 clinic visits 2 ., These volunteers make their contributions with the knowledge that the scientific advancements gained may not benefit them ., Therefore calling for frequent testing—which from a modeling perspective would be ideal—is problematic ., However we note , due to the lack of data in the first week following ATI , that we made simplifying assumptions , such as neglecting ART decay kinetics in study participants whose pre-ATI ART regimen excluded NNRTIs ., Therefore we call for more regular data collection in the 1-2 weeks if possible , which may be particularly illuminating in characterizing rapid viral rebound and thus improving parameter estimation , without over-burdening clinical trialists or generous volunteers ., In our modeling we neglected the inherent heterogeneity of the latent reservoir and associated latent cell activation rates ., Latently infected cells are in majority memory cells 45 , each of which may be specific for a pathogen or set of pathogens ., There is evidence that the reservoir is composed of clonal populations 53 , 60 , 75 , so there may be genetically homogeneous subsets of cells , but even cells in clones may exhibit differing activation dynamics ., However , in modeling viral rebound for large latent reservoir sizes , we neglect this heterogeneity in favor of the mean activation rate ., Heterogeneity in activation rate becomes important as latent reservoir sizes gets small , and we move towards elimination , but that is not the focus of this present study ., The latent cell activation rate , a , is part of the recrudescence rate , as ( 1 − q0 ) , which we estimate from the data and which we assume to be the same for all individuals ., Note that we can only estimate the parameter combination as ( 1 − q0 ) , and not a itself ., Future modeling efforts may aim to rectify this possibly by including additional data , such as direct estimates of the pre-ATI viral reservoir size and viral growth rates and viral set points post-ATI for each study participant ., In accounting for the delay in viral rebound observed in Li et al . ( 2016 ) associated with inclusion of NNRTIs in the pre-ATI ART regimen 2 , we assumed that the probability that latent cell activation induces viral rebound decays expontentially to q0 at rate k ., Our estimate for k , which should account for the rate of drug decay post-ATI , was at least order of magnitude lower than the mean NNRTI concentration decay rates in plasma 55 ., One explanation is that most T-cells and latently infected cells reside in lymphatic tissues 76 , 77 ., The pharmacokinetics and pharmacodynamics in the lymphatic tissues are not clear , although there is evidence that drug penetration is lower 78 , 79; commensurately , drug clearance may also be slower ., Since our model crudely treats the whole body as homogeneous , the expression for decaying effectiveness of the drug must average the dynamics in different tissues , potentially explaining the inconsistencies ., It is also interesting to note that the NNRTI efavirenz’s clearance is dependent on CYP450 2B6 gene polymorphism and there are certain polymorphisms that increase plasma half life such that drug levels above the 95% inhibitory concentration maybe present for > 21 days after treatment interruption 48 ., However , we acknowledge that our estimates of the rate at which drug loses effectiveness will need to be further refined and validated ., In future studies we will attempt to refine q ( t ) and more carefully address its variability , potentially , by considering specific drug pharmacokinetic/pharmacodynamics in tissues , when available , and data on viral dynamics post-rebound to inform the reproductive ratio in absence of ART ., It is also possible that ART decay may occur in a biphasic manner with our estimate of k reflecting the terminal elimination phase ., Additional pharmacokinetic studies are needed to explore this possibility ., We also simplified our model by taking a constant reservoir size , neglecting factors contributing to long-term reservoir decay such as latent cell death and proliferation 45 ., Preceding viral rebound , we anticipate that the latent reservoir would continue to decay at on-therapy rates 43 , 80 resulting from these dynamics ., But the reservoir is typically large and its decay is slow , with a 44 month half-life on average 43 , 80 , so in our 60-day rebound period , the average reservoir size would decrease by less than 3% ., Thus our constant reservoir size assumption is reasonable ., We can extend our simple model to include latent cell proliferation and death , derived in the S1 Text ., Intriguingly , the resulting expression for probability of viral rebound gives the natural activation rate a as , in principle , an identifiable parameter , in contrast to our simple model , for which only the successful latent cell activation rate , ( 1 − q ) aL0 , is identifiable ., Unfortunately , to disentangle a from other parameters in the extended model , we require more refined data , as discussed based on the mathematics in the S1 Text ., Such data may be difficult to obtain; as it is , the data from Li et al . 2 which we employ is the most extensive and well-curated ATI study data currently available ., In integrating study participant data into our viral rebound model , we made the assumption that log10 ( cell-associated HIV RNA ) is proportional to the size of the replication-competent portion of the latent reservoir | Introduction, Materials and methods, Results, Discussion | Antiretroviral therapy ( ART ) effectively controls HIV infection , suppressing HIV viral loads ., Suspension of therapy is followed by rebound of viral loads to high , pre-therapy levels ., However , there is significant heterogeneity in speed of rebound , with some rebounds occurring within days , weeks , or sometimes years ., We present a stochastic mathematical model to gain insight into these post-treatment dynamics , specifically characterizing the dynamics of short term viral rebounds ( ≤ 60 days ) ., Li et al . ( 2016 ) report that the size of the expressed HIV reservoir , i . e . , cell-associated HIV RNA levels , and drug regimen correlate with the time between ART suspension and viral rebound to detectable levels ., We incorporate this information and viral rebound times to parametrize our model ., We then investigate insights offered by our model into the underlying dynamics of the latent reservoir ., In particular , we refine previous estimates of viral recrudescence after ART interruption by accounting for heterogeneity in infection rebound dynamics , and determine a recrudescence rate of once every 2-4 days ., Our parametrized model can be used to aid in design of clinical trials to study viral dynamics following analytic treatment interruption ., We show how to derive informative personalized testing frequencies from our model and offer a proof-of-concept example ., Our results represent first steps towards a model that can make predictions on a person living with HIV ( PLWH ) ’s rebound time distribution based on biomarkers , and help identify PLWH with long viral rebound delays . | Antiretroviral therapy ( ART ) effectively controls HIV infection , holding HIV viral loads to levels undetectable by commercial assays ., Therapy interruption is followed by rebound of viral loads to high , pre-therapy levels , but there is significant heterogeneity in the timing of the rebound to those high levels ., Some rebounds occur within days , weeks , or even , rarely , years ., Here we develop a mathematical model to characterize rebounds occurring within two months of treatment interruption ., Li et al . ( 2016 ) report biological markers that correlate with the time between ART interruption and viral rebound ., We incorporate this information to parametrize our model so that our model can make predictions on time to rebound tailored to the individual undergoing ATI ., Our parametrized model can aid in design of clinical trials to study infection dynamics following treatment interruption ., We also use our model to gain insight into the underlying within-host viral dynamics ., For example , we refine previous estimates of viral recrudescence after ART interruption and determine a recrudescence rate of once every 2-4 days ., Our results represent first steps towards a model that can make predictions on an person living with HIV’s rebound time based on personal biomarkers , and help identify patients with long viral rebound delays . | cell physiology, medicine and health sciences, pathology and laboratory medicine, viral transmission and infection, cell activation, antiviral therapy, pathogens, immunology, microbiology, biomarkers, retroviruses, viruses, immunodeficiency viruses, preventive medicine, mathematics, rna viruses, antiretroviral therapy, vaccination and immunization, viral load, public and occupational health, infectious diseases, probability density, medical microbiology, hiv, microbial pathogens, viral replication, probability theory, biochemistry, viremia, cell biology, virology, viral pathogens, biology and life sciences, viral diseases, physical sciences, lentivirus, organisms | null |
journal.pntd.0004232 | 2,015 | Comparative Genomics of Field Isolates of Mycobacterium bovis and M. caprae Provides Evidence for Possible Correlates with Bacterial Viability and Virulence | Mycobacterium tuberculosis has infected more than 2 . 5 billion people worldwide with approximately 9 million new tuberculosis ( TB ) cases reported every year 1 ., Animal TB is caused by infection with Mycobacterium bovis and closely related members of the M . tuberculosis complex ( MTBC ) such as M . caprae ., Although cattle are the main concern regarding animal TB in industrialized countries , several other mammals including humans are also infected 2 , 3 ., Eurasian wild boar ( Sus scrofa ) are a natural reservoir for M . bovis in some regions and thus vaccination strategies are being developed for TB control in this species 4–8 ., Several other domestic and wild animals are also infected with M . bovis and may act as reservoir species 9–13 ., The life cycle of mycobacteria is complex and the mechanisms resulting in pathogen infection and survival in host cells are not fully understood 14 ., Nevertheless , it is generally accepted that after inhalation into the lung or entry to the oropharyngeal cavity , the principal entry routes , mycobacteria of the MTBC are phagocytized by macrophages , which constitute the main host cell ., As with other intracellular bacteria , mycobacteria survive inside macrophages by escaping host immune response , which results in the formation of a granuloma that effectively contains infected cells ., A change in the host-bacterial equilibrium of granulomas is thought to result in the release of infected cells outside containment and onward transmission of mycobacteria to susceptible hosts 14 ., The association between M . bovis spoligotypes and TB lesions in cattle has been used to correlate bacterial genotype with virulence 15 ., However , these genotyping methods cover only a small portion of the approximately 4 , 000 genes contained in the 4 . 4 Mb mycobacterial genome 16 ., Recently , whole-genome sequencing and comparative genomics analyses have provided new insights into the evolution and adaptation of the MBTC to survive inside the host and explained phenotypic traits related with transmissibility and virulence 16–21 ., Although the first M . tuberculosis genome sequence was reported in 1998 16 , it was not until 2003 when the first M . bovis genome was sequenced 19 ., Presently , a large number of M . tuberculosis but few M . bovis ( except for BCG strains ) genome sequences are available 18 ., The relative paucity of M . bovis genome sequence information limits the possibility of characterizing mycobacterial evolution and correlation with virulence at the whole-genome level ., In this study , the genome of three M . bovis ( MB1 , MB3 , MB4 ) and one M . caprae ( MB2 ) field isolates with different lesion score , prevalence and host distribution phenotypes were sequenced ., Genome sequence information was used for whole-genome and protein-targeted comparative genomics analysis with the aim of finding correlates of phenotypic variation with potential implications for TB disease risk assessment and control ., All animal sampling was post-mortem ., Wildlife samples came from hunter-harvested individuals that were shot during the legal hunting season independently and prior to our research while livestock samples were obtained at the slaughterhouse where they were being processed as part of the normal work and submitted to the reference laboratory by the slaughterhouse veterinarian ., According to EU and National legislation ( 2010/63/UE Directive and Spanish Royal Decree 53/2013 ) and to the University of Castilla–La Mancha guidelines no permission or consent is required for conducting this type of study ., Field isolates used come from the EU Reference Laboratory for Animal Tuberculosis ( VISAVET ) ., Three M . bovis ( MB1 , MB3 , MB4 ) and one M . caprae ( MB2 ) field isolates were selected for this study ( Table 1 ) ., These isolates were originally obtained from wild boar ( MB3 , MB4 ) , cattle ( MB1 ) and goat ( MB2 ) ., The study focused on Ciudad Real Province , Spain ., This is a high ungulate density area , the west side of the province composed by interspersed game ranges and protected nature areas , with persistent TB infection in extensive livestock farms 22 ., Nine hundred MTBC isolates collected from wild ungulates and livestock from 2000 to 2011 were spoligotyped resulting in 62 different spoligotypes ( 23 and S1 Fig ) ., The criteria for selection of the MTBC spoligotypes included in the study were based on: Following these three criteria , the M . bovis spoligotypes SB0339 ( MB1 , MB4 ) and SB0134 ( MB3 ) were selected ., The M . caprae spoligotype SB0157 ( MB2 ) was included in the study as an outgroup but closely related species 29 and for the increased proportion of M . caprae isolated from bovine samples during 2004–2009 30 ., The MB4 isolate was included because although it has the same spoligotype as MB1 , it served as a model to characterize possible differences between MTBC isolates otherwise classified with the same spoligotype ., The four isolates were grown in 15 ml of Middlebrook 7H9 liquid media supplemented with 0 . 36% sodium pyruvate and 10% OADC ( Oleic Albumin Dextrose Catalase ) for 5 weeks ., Chromosomal DNA samples were obtained as described by van Soolingen et al . 31 ., Briefly , cultures were centrifuged and pellets were washed twice in 5 ml water ., Mycobacteria were heated at 100°C for 15 min to kill the cells ., After centrifugation , the cells were resuspended in 5 ml TE buffer ( 0 . 01 M Tris-HCl , 0 . 001 M EDTA , pH 8 . 0 ) ., Lysozyme was added to a final concentration of 1 mg/ml and the tube was incubated over night at 37°C ., Eighty hundred and seventy-five microliters of 10% sodium dodecyl sulfate ( SDS ) with 62 . 5 μl of proteinase K ( at a 10-mg/ml concentration ) were added , and the mixture was incubated for 1 h at 60°C ., The extract was transferred to a phase lock gel tube ( prime5 , Fisher Scientific SL , Madrid , Spain ) for a phenol/chloroform DNA extraction ., Genomic DNA ( 2–5 μg ) was subjected to mechanical fragmentation using a BioRuptor ( Life Technologies , Carlsbad , CA , USA ) ., The number of cycles was adjusted to obtain DNA fragments of a final average size of about 500 pb ., Samples were then used to prepare sequencing-amenable TruSeq libraries ( NEB-Next , New England Biolabs , Ipswich , MA , USA ) ., Briefly , DNA fragments were made blunt-ended , phosphorylated , adenylated and Illumina-compatible adapters were ligated ., After purification , barcoded sequences as well as Illumina-specific sequences were introduced by PCR , followed by quantitation of individual libraries ., Libraries were then pooled and quantified again ., A quality control of the pooled library made in bioanalyzer is shown in the figure , including an estimation of the percentage of non-overlapping reads that could be obtained using a 2x250 paired-end sequencing protocol ., Library was qPCR-quantitated and brought to a final concentration of 10 nM ., DNA was then denatured and equilibrated so that a final concentration of 18 pM of library was loaded onto a MiSeq v . 3 flowcell ( Illumina , San Diego , CA , USA ) and sequenced using a 2x250 paired-end sequencing protocol to obtain more than 400x high quality coverage ( 1 . 9–2 . 5 Gb ) with 84% of the bases showing a Q30 factor > 30 ., Reads were finally split according to barcodes and used for bioinformatics analysis ., High quality overlapping reads were merged using FLASH ( Magoc et al . , 2011 ) and then assembled using Velvet 32 with k-value = 97 ( S1 Table ) ., Contigs were annotated using BG7 33 , 34 ( S2 Table ) ., For annotation , a set of 191 , 017 reference proteins was used including, ( a ) all Uniprot proteins from M . bovis and M . tuberculosis ,, ( b ) a set of bacterial antibiotic resistance related Uniprot proteins selected using the GO annotation terms “antibiotic resistance” and “response to antibiotic” and a selection of proteins based on similarity to the proteins of ARDB 35 , and, ( c ) all Uniprot proteins with Enzyme Code ( EC ) from MTBC ., For whole genome comparative analysis , the 4 genomes were then aligned against the M . bovis reference genome sequence ( AF2122/97; http://www . ncbi . nlm . nih . gov/nuccore/31791177 ) using Differences program ( Turrientes et al . , 2010 ) that allows comparisons at the whole genome level and particularly the detection of substitutions , insertions or deletions of any length and at any region of the genome ., Genome sequence information and annotation was deposited in GenBank under the accession numbers CDHF01000001-CDHF01000049 , CDHG01000001-CDHG01000059 , CDHH01000001-CDHH01000094 and CDHE01000001-CDHE01000118 for MB1-MB4 isolates , respectively 36 ., A phylogenetic tree was constructed based on the SNPs found at the core genome sequence shared with a similarity over a threshold between all genomes included in the analysis using Harvest 37 and visualized with EvolView ( v 198 . 3 ) 38 ., Harvest defines the SNPs aligning whole assembled genomes and not reads like in other approaches , thus allowing the identification of the gene or the intergenic region where differences are allocated ., The SNPs are included in the . vcf file provided by Harvest suite ., The genomes used in the SNP phylogenetic analysis using Harvest included MB1-MB4 , M . bovis AF2122/97 ( http://www . ncbi . nlm . nih . gov/nuccore/31791177 ) , M . bovis ATCC BAA-935 ( http://www . ncbi . nlm . nih . gov/nuccore/690294709 ) , M . bovis BCG Pasteur 1173P2 ( http://www . ncbi . nlm . nih . gov/nuccore/121635883 ) , and M . tuberculosis H37Rv ( http://www . ncbi . nlm . nih . gov/nuccore/448814763 ) ., The multilocus sequence analysis was conducted using 18 genes coding for the proteins linking stress response with lipid metabolism ( S3 Table ) ., The nucleotide sequences of the genes were obtained from the genomes of MB1-MB4 isolates used in this study ., For comparison , the same sequences were obtained from the reference M . bovis BCG Pasteur 1173P2 , M . bovis AF2122/97 and M . tuberculosis H37Rv ., M . canettii ( NCBI reference sequence NC_019950 ) was used as outgroup ., The nucleotide sequences were concatenated and then aligned with MAFFT ( v7 ) , configured for the highest accuracy 39 ., After alignment , regions with gaps were removed and 20080 gap-free sites were used in maximum likelihood phylogenetic analysis as implemented in PhyML ( v3 . 0 aLRT ) 40 , 41 ., The reliability for the internal branches was assessed using the approximate likelihood ratio test ( aLRT–SH-Like ) 41 ., Graphical representation and editing of the phylogenetic trees were performed with EvolView ( v 198 . 3 ) 38 ., The alignments obtained by MAFFT were used to perform codon alignments using HIV database website ( www . hiv . lanl . gov; 42 ) ., Non-synonymous ( dN ) and synonymous ( dS ) nucleotide substitutions were classified using the SNAP method 43 implemented in HIV database website 42 ., SNPs were identified by pairwise comparison of MB1-MB4 and M . bovis AF2122/97 using SNAP ., The dN/dS ratio was calculated for the 18 genes coding for the proteins linking stress response with lipid metabolism ( S3 Table ) and for 81 antigen-coding genes ( S4 Table ) present on each of the MB1-MB4 isolates included in the study ., Under the Datamonkey server ( http://www . datamonkey . org; 44 ) , the algorithm SLAC 45 was used to detect which nucleotide substitution site were positively or negatively selected ., For each dS and non-synonymous dN substitution site , four measurements were made: normalized expected ( ES and EN ) and observed numbers ( NS and NN ) ., The SLAC algorithm then calculated: dN = NN/EN and dS = NS/ES ., If dN < dS a codon was negatively selected and if dN > dS a codon was positively selected ., A two-tailed extended binomial distribution set at P<0 . 05 was used to assess significance of the algorithm ., The SLAC algorithm uses a neighbor-joining tree with a maximum likelihood for branch lengths and substitution rates ., To identify non-synonymous mutations that may be associated with virulence and/or transmission , the orthologs of peptidoglycan assembly protein locus Rv0050 ( H37Rv ) in strains MB1-MB4 were compared to the equivalent locus in animal isolates ( BCG Pasteur 1173P2 , AF2122/97 , 09–1192 ) and human isolates ( Bz 3115 , B2 7505 ) of M . bovis available in the GenBank ., The M . bovis strains were selected among all strains for having distinct Rv0050 locus and/or distinct source of isolation ., The presence of a putative signal peptide and cleavage site in the Rv0050 locus was analyzed with a previously validated program for their prediction in M . tuberculosis ( SignalP; http://www . cbs . dtu . dk/services/SignalP-3 . 0/ ) 46 ., Results were also confirmed with two other programs , Signal-Blast ( http://sigpep . services . came . sbg . ac . at/signalblast . html ) and Phobious ( http://phobius . sbc . su . se/ ) ., To confirm selected SNPs identified in the mycobacteria genomes sequenced in this study , sequence-specific oligonucleotide primers were design using reference genomes M . bovis AF2122/97 ( BX248333 . 1 ) or M . tuberculosis H37Rv ( AL123456 . 3 ) for PCR and sequencing of the amplicons ., Selected loci and direct and reverse primers used for analysis included Rv0050 ( ponA1; 5´-GACTTTCCCCAAACCGACCGAGG-3’ and 5’-GATCGGTCCCCCGACCACCATT-3’ ) , Rv0589 ( MCE2a; 5´-GTGCCAACGCTGGTGACGAG-3’ and 5’-AGAACACGATCAACCCATGA-3’ ) , Rv1198 ( ESAT-6; 5´-ATGACCATCAACTATCAATT-3’ and 5’-TCGGCTCCAGCTGGGCCTGA-3’ ) , and Rv1860 ( FAP-B; 5´-ATGCATCAGGTGGACCCCAA-3’ and 5’- AGCGGACCTTACCGGCCTGA-3’ ) ., The PCR was conducted using 2 μl of DNA with 20 pmol of each primer in a 50 μl reaction PCR Master Mix ( Promega , Madison , WI , USA ) using a GeneAmp PCR System 2700 thermocycler ( Applied Biosystems , Carlsbad , CA , USA ) ., PCR products were electrophoresed on 1 . 5% agarose gels to check the size of the amplified fragments by comparison to molecular weight marker GeneRuler 1kb DNA ladder ( Thermo Scientific , Waltham , MA , USA ) ., Amplified DNA fragments were purified with a PureLink Quick PCR Purification Kit ( Thermo Scientific , Waltham , MA , USA ) and sequenced using the reverse primer on each locus ., Amplicons from at least two independent PCR reactions were sequenced ., Among the 900 MTBC field isolates analyzed , 62 different spoligotypes were identified suggesting that the study area is one of the regions with the highest diversity of MTBC spoligotypes described in the literature 47 ., The high genetic diversity of MTBC in this area supported an important natural scenario where MTBC and particularly M . bovis have diversified , thus offering an interesting epidemiological and evolutionary context where new genotypes can emerge and diversify in terms of adaption to host under a range of environmental and human driven factors ., The spoligotypes SB0121 , SB0134 , SB0339 , SB0120 , SB1263 and SB0157 were among the most frequent ones found in both livestock and wild ungulates in the study area ( S1 Fig ) ., The spoligotypes found at a higher frequency than expected given a predicted mutation rate were identified ( Fig 1A ) and represented in a hierarchical tree ( Fig 1B ) ., The spoligotypes SB0134 ( MB3 ) and SB0339 ( MB1 and MB4 ) , which clustered in space and time and showed low mutation rates but high abundance , were selected as emergent under study conditions ., The output hierarchical tree suggested a history of mutation events and a relationship between these spoligotypes with different levels of delection for spoligotypes SB0134 ( MB3 ) and SB0339 ( MB1 and MB4 ) ( Fig 1B ) ., These spoligotypes were therefore chosen for this study as fulfilling selection criterion, ( i ) and, ( ii ) described above ., Additionally , findings in Iberian red deer showed a higher lesion score caused by spoligotype SB0134 when compared to spoligotype SB0339 ( Mann-Whitney U test; p = 0 . 04 ) while these spoligotypes and particularly SB0134 and SB0157 suggested an association with high TB severity in Eurasian wild boar thus also fulfilling selection criteria, ( iii ) ( Fig 1C ) ., In summary , three M . bovis ( MB1 , MB3 , MB4 ) and one M . caprae ( MB2 ) field isolates with different prevalence , lesion score and host distribution phenotypes were selected for this study ( Table 1 ) ., MB3 showed high distribution and lesion score while MB1 and MB4 were highly distributed but with low lesion score ., The M . caprae ( MB2 ) isolate had moderate distribution and high lesion score and was selected for comparison with the M . bovis isolates ., These phenotypic variations are relevant for pathogen transmission and virulence and could be correlated with genome sequence information with implications for TB disease risk assessment and control ., The results of the phylogenetic analysis showed that the MB1 and MB4 isolates with the same spoligotype were the most closely related isolates ( Figs 2A and S2 ) ., The MB2 isolate ( M . caprae ) clustered separately from the other isolates , which clustered together with M . bovis sequences ., Nevertheless , M . caprae ( MB2 ) was closely related to M . bovis when compared to M . tuberculosis ( Fig 2A ) ., The genome of M . bovis BCG Pasteur 1173P2 was the most similar to all sequenced mycobacteria genomes ( Figs 2A , 2B and 3 ) ., It has been proposed that during evolution , a clone of M . tuberculosis that was originally adapted to cause human TB evolved to infect a non-human mammal and thus began the transition into non-human ecotypes such as M . bovis , which in turn spread to cattle , goats , oryx , seals and pigs 20 ., Our results supported a close relationship between M . bovis isolates and suggested that M . caprae is one of the M . bovis-related mycobacteria adapted to infect goats and sheep as well as other hosts such as wild boar , red deer , cattle and humans 29 , 48 , 49 ., To better understand the relationship between these isolates , a comparative genomics approach was used ., The results showed the presence of translocations , deletions of small genomic regions and SNPs between genomic sequences ( Figs 2B , 3 and S2 ) ., Large-scale polymorphism studies have demonstrated that the MTBC shows a large number of deletions of small genomic regions consistent with the reductive evolution typical of intracellular bacteria 20 ., Nevertheless , sequential chromosomal nucleotide substitutions are considered to be the main driver in the M . tuberculosis genome evolution 20 ., The results reported here supported these findings for M . bovis isolates and suggested that the isolates with high lesion score , MB2 and MB3 , contained the largest number of polymorphisms when compared to the MB1 and MB4 isolates with low lesion score ( Figs 2B and 3 ) ., However , a clear correlation between phenotype and genome sequence requires a protein-targeted comparative analysis between the different isolates ., For protein-targeted analysis , the study was focused on proteins that are known to play an important role in mycobacterial viability or virulence ., Protein-targeted comparative analysis was conducted for, ( a ) the ESX or type VII secretion system ,, ( b ) proteins linking stress response with lipid metabolism ,, ( c ) host T cell epitopes of mycobacteria ,, ( d ) antigens and, ( e ) peptidoglycan assembly protein to define possible correlates with bacterial virulence and viability or distribution . | Introduction, Materials and Methods, Results and Discussion | Mycobacteria of the Mycobacterium tuberculosis complex ( MTBC ) greatly affect humans and animals worldwide ., The life cycle of mycobacteria is complex and the mechanisms resulting in pathogen infection and survival in host cells are not fully understood ., Recently , comparative genomics analyses have provided new insights into the evolution and adaptation of the MTBC to survive inside the host ., However , most of this information has been obtained using M . tuberculosis but not other members of the MTBC such as M . bovis and M . caprae ., In this study , the genome of three M . bovis ( MB1 , MB3 , MB4 ) and one M . caprae ( MB2 ) field isolates with different lesion score , prevalence and host distribution phenotypes were sequenced ., Genome sequence information was used for whole-genome and protein-targeted comparative genomics analysis with the aim of finding correlates with phenotypic variation with potential implications for tuberculosis ( TB ) disease risk assessment and control ., At the whole-genome level the results of the first comparative genomics study of field isolates of M . bovis including M . caprae showed that as previously reported for M . tuberculosis , sequential chromosomal nucleotide substitutions were the main driver of the M . bovis genome evolution ., The phylogenetic analysis provided a strong support for the M . bovis/M ., caprae clade , but supported M . caprae as a separate species ., The comparison of the MB1 and MB4 isolates revealed differences in genome sequence , including gene families that are important for bacterial infection and transmission , thus highlighting differences with functional implications between isolates otherwise classified with the same spoligotype ., Strategic protein-targeted analysis using the ESX or type VII secretion system , proteins linking stress response with lipid metabolism , host T cell epitopes of mycobacteria , antigens and peptidoglycan assembly protein identified new genetic markers and candidate vaccine antigens that warrant further study to develop tools to evaluate risks for TB disease caused by M . bovis/M . caprae and for TB control in humans and animals . | Mycobacteria belonging to the Mycobacterium tuberculosis complex infect humans and animals since pre-history and are a serious health problem worldwide ., Whole-genome sequencing and comparative genomics generate information on the evolution and molecular basis of pathogenicity and transmissibility ., However , while genomic information is increasingly available for the main human pathogens such as Mycobacterium tuberculosis , little is known about closely related bacteria , Mycobacterium bovis and Mycobacterium caprae ., These mycobacteria infect humans causing zoonotic tuberculosis and are the main causative agents of animal tuberculosis ., Although human-to-human transmission of zoonotic tuberculosis is limited , the infection often causes extra-pulmonary disease in humans and is still a major public health concern in developing countries , causing not only human disease but also severe effects on livelihoods ., In this study , whole-genome sequences and targeted comparative genomics of three Mycobacterium bovis and one Mycobacterium caprae field isolates generated new information on the evolution and phenotypic variation of these mycobacteria ., The results identified new genetic markers and candidate vaccine antigens that warrant further study to develop tools to evaluate risks for tuberculosis caused by M . bovis/M . caprae and for disease control in humans and animals . | null | null |
journal.ppat.1003023 | 2,012 | Trypanosome Motion Represents an Adaptation to the Crowded Environment of the Vertebrate Bloodstream | Blood vessels form a dense network throughout the human body with a total length of about 100 , 000 kilometers ., The vessels diameter ranges from a few micrometers in capillaries to centimeters in the aorta and veins ., Blood contains about 45% ( v/v ) cellular components , which flow with velocities ranging from mm s−1 in capillaries to m s−1 in the aorta ., Viscous forces and laminar flow are dominant in blood circulation ., In small capillaries , red blood cells ( RBC ) move in a single row , while in larger vessels they are thought to accumulate in the channel center due to hydrodynamic flow effects ., Despite these fundamental characteristics , blood composition , temperature , pressure and oxygen content differ significantly between vertebrate species ., Nevertheless , the parasitic unicellular trypanosomes prosper in the circulation of all vertebrate classes , from fish to bird ., Thus , the parasites have evolved by adapting to very different bloodstream conditions ., Some trypanosome species cause deadly diseases in livestock and man , e . g . the African sleeping sickness ., Human African Trypanosomiasis ( HAT ) is an exemplary disease of poverty ., There are only very few and rather ancient drugs available , which in addition are highly toxic ., Most critically , in many sub-Saharan countries , health agencies have essentially lost control of HAT due to social and geopolitical problems; consequential poor public health implementation has resulted in widespread emergence of drug resistance ., Thus , new medication is urgently needed ., Unraveling the unique cellular and molecular features that distinguish trypanosomes from other eukaryotes has been a prime goal in the search for promising drug targets , however , success has been limited so far ., An alternative approach appears to be to study the behavior of trypanosomes in their natural environment , namely the mammalian bloodstream , where the cells are constantly opsonized with antibodies and serum factors ., The only barrier that shields trypanosomes from the host is an astonishingly dense cell surface coat , which is made of 107 copies of the same type of lipid-anchored , variant surface glycoprotein ( VSG ) ., The trypanosomes use a system of antigenic variation to evade the hosts immune response , whereby they randomly switch the exposed variant glycoproteins ( VSG ) and thus escape detection 1 ., Due to the large number of structurally related but immunologically distinct VSGs , vaccination against HAT appears impossible ., However , a second immune evasion strategy could prove more versatile for intervention as it directly involves cellular motility , which is thought to be essential for the parasites 2–4 ., Bloodstream form ( BSF ) trypanosomes swim to rapidly remove surface-bound antibodies 5 , 6 ., For propulsion , they utilize a single leading flagellum , which emerges from the flagellar pocket , follows the cell body , to which it is attached , and protrudes freely at the anterior of the cell ( Fig . 1A ) ., The motion relative to the surrounding fluid generates a hydrodynamic drag , which causes antibody-bound VSGs to drift in the plane of the plasma membrane towards the trailing posterior end and into the flagellar pocket 5 ., This invagination of the plasma membrane is the only place where endocytosis , an extremely fast process in trypanosomes , takes place 7 ., The mechanism of hydrodynamic protein sorting greatly accelerates the diffusion-limited uptake of host antibodies , provided that trypanosomes exhibit fast directional motion ., The aim of the present study was to elucidate in detail the biomechanics of trypanosome motility and relate this to the parasites life in blood ., The swimming behavior of cultivated trypanosomes appears to be highly variable , with few cells showing persistent directional motion ( Table 1 ) ., Persistence is defined here by continuous directional movement for at least 150 micrometers , whereas non-persistent motion is characterized by shorter swimming trajectories interrupted by tumbling phases , during which the trypanosome stops translocation and changes its orientation 8 , 9 ., In cell culture , the measured mean population velocity was relatively slow ( ( 5 . 7±0 . 11 ) µm s−1; SEM , n\u200a=\u200a979 ) ., This is slower than the velocity required for efficient removal of host antibodies 5 ., In contrast , in blood the cells reached much higher velocities of 30 µm s−1 ., Obviously , the physicochemical conditions in blood and cell culture differ greatly ., Therefore , we systematically analyzed the motility of trypanosomes in varying environments ., Possible factors that influence swimming behavior include chemical cues , oxygen content , pressure , viscosity , flow , confinement and presence of blood cells ., In blood freshly drawn from infected mice , virtually all cells swam persistently ( Table 1 ) ., In contrast , less than a third of cells were persistent swimmers in standard cell culture medium and velocities varied significantly ., We found no difference in the swimming behavior of trypanosomes in blood of pre-infected and uninfected animals , ruling out that the parasites secrete a motility-promoting factor ., Blood plasma or serum only marginally influenced the percentage of mobile cells ( Table 1 ) ., Thus , chemical cues are unlikely to affect trypanosome motility in the bloodstream ., We also observed the reduction of oxygen partial pressure to have no influence on motility ., However , when we raised the viscosity of the cell culture medium , the number of persistent swimmers increased , as well as the velocity of trypanosomes ( Fig . 1B ) ., Upon addition of 0 . 4% ( w/v ) methylcellulose the viscosity of the medium equals that of blood ., Under these conditions the percentage of persistently swimming cells doubled and the mean population velocity almost tripled ( Fig . 1B–D ) ., Trypanosomes were capable swimmers even when viscosity was raised to 4000 mPa s , three orders of magnitude higher than that of blood ., This compares to mammalian spermatozoa , which are also adapted to move in a broad viscosity range ., In sperm , the flagellar beat frequency is decreased in high viscosity medium , but the progressive velocity does not change , as kinetic efficiency rises 10 , 11 ., This probably involves the interaction with microstructures contained in the methylcellulose solution 11 ., The forces trypanosomes would need to exert in order to move through homogenous fluids of such high viscosity are significant ., We assume that the increase in velocity , without the need to produce such high forces , can be explained by methylcellulose forming loose and quasi-rigid networks consisting of long , linear polymer molecules ., This was suggested first by Berg and Turner in 1979 for bacteria , cilia and flagella 12 and subsequently mathematically developed for certain bacteria 13 ., As these bacteria do , small organisms like trypanosomes , even though 3 to 4 times larger in diameter , also seem to be able to wriggle through these networks without experiencing the resistive force of the apparent macroscopic viscosity ., We have estimated the minimal force that the flagellum generates to be 5 pN by bead displacement experiments ( Video S1 ) ., We also found that trypanosomes produced a force of maximally 100 pN , as they were not able to bend the various poly-dimethyl siloxane ( PDMS ) -pillars used in our experiments ., The force required for bending these pillars is similar to that needed for the deformation of RBC ., This agrees with the fact that although RBC can be readily displaced by trypanosomes , erythrocytes cannot be deformed by the parasite ( Video S1 ) ., Thus , the force exerted by the flagellum is at least one order of magnitude lower than reported previously 14; the most likely numbers have recently been measured by optical trapping experiments and are below 10 pN 15 ., In fact , for swimming , trypanosomes do not require larger forces; the absolute value of the propulsive force that has to be generated equals the viscous drag force acting in the opposite direction ., The drag force F can be calculated by the Stokes equation , assuming the cell body to be a sphere with known radius: ( 1 ) where η is the dynamic viscosity of the medium , r is the radius of the sphere and υ is the swimming velocity ., For a trypanosome ( r\u200a=\u200a1 . 5 µm ) swimming in cell culture medium ( η\u200a=\u200a0 . 95 mPa s ) at a velocity υ of 20 µm s−1 , the viscous drag force is 0 . 54 pN ., The viscosity of blood directly depends on the concentration of cellular components 16 , 17 and the presence of particles renders blood a non-Newtonian fluid , meaning that its viscosity changes with flow speed ., This fact , in addition to continuous , vigorous self-mixing and formation of erythrocyte stacks ( ‘rouleaux’ ) , makes it difficult to measure cell motility in blood directly 18 ., Hence , we aimed to quantify the behavior of trypanosomes in environments that resemble aspects of blood physics , but are defined and more easily manipulable ., Initially , we chose suspensions of different kinds of particles that produce viscous fluids similar to blood ., We generally observed enhanced cell motilities in the presence of particles , independent of their size , shape and mass ., Live and chemically fixed erythrocytes influenced trypanosome motion in a manner similar to polystyrene and metal beads or nanodiamonds ( Video S1 ) ., However , blood is a self-mixing system , meaning that the position of cells is continuously randomized due to hydrodynamic flow under confinement ., In order to simulate this complex situation we used continuous flow microfluidics , which to date however , is not compatible with high-resolution , quantitative microscopic imaging of fast moving objects in three dimensions ., Therefore , we devised a scenario that resembles a “frozen” suspension ., Trypanosome motility was measured in arrays of regularly aligned inert PDMS pillars in order to simulate the crowded environment of the bloodstream ( Fig . 2A ) ., We observed a striking increase in the percentage of persistently swimming cells and higher velocities for cells mechanically interacting with pillars ( Fig . 2B ) ., About 90% of trypanosomes exhibited a fast and directional motion in arrays of pillars whose size ( 8 µm ) and regular spacing ( 4 µm ) was comparable to RBC in blood ( Fig . 2C , D ) ., A maximum speed of 40 µm s−1 was measured within these arrays , which is almost 8-times faster than the mean population velocity in cell culture ., No difference was found when the experiments were conducted in trypanosome dilution buffer ( TDB ) instead of culture medium , ruling out that compounds from fetal calf serum were involved ., We conclude that objects with the size and spacing of RBC are required and sufficient to promote maximum directional forward velocity of trypanosomes ( Table, 2 ) and consequently , the effectual removal of cell-surface bound host antibodies ., Noteworthy , in the presence of narrower spaced pillar arrays , as well as in artificial collagen networks , around half of the persistent swimmers were observed to move backwards , i . e . with the flagellum trailing ., This behavior had never been observed for wild type cells , but only for motility mutants with the outer dynein arms of the axoneme missing 5 , 19 ., It was tempting to speculate that pure mechanic resistance in close-meshed environments could initiate a simple but faultless trap escape mechanism by triggering trypanosome backward motion ., In order to understand the complex swimming behavior , we detailed the mechanism of movement for trypanosomes swimming either forwards , backwards or tumbling ., High-speed transmitted light microscopy imaging allows analysis of cell motility with sufficient resolution in space and time 20 ., Here , we introduce fluorescence microscopy utilizing sCMOS technology , which combines kHz frame rates with very high sensitivity ., High-speed fluorescence microscopy has the dual advantage of providing defined out-of-focus information as well as high-resolution structural content , allowing the derivation of three-dimensional information at unprecedented temporal resolution ., Using fluorescence surface labeling and high speed imaging , we observed the exact course of the flagellum and the cell body around individual pillars during successive flagellar beats , unequivocally demonstrating the adaptation of cell morphology to the surrounding erythrocyte-sized obstacles ( Fig . 2D ) ., Only forward swimming cells whose flagellar wavelength and amplitude , as well as the resulting dynamic curvature of the cell body matched the spacing between PDMS-pillars achieved maximum directional velocity ., These cells exhibited continuous tip-to-base beats of the flagellum at a regular frequency of ( 18 . 3±2 . 5 Hz ) ., The amplitude of the free flagellar tip , defined as half the maximum transverse displacement of the flagellum relative to the anterior-posterior cell axis , was ( 3 . 3±0 . 7 µm ) ., This specifically allowed mechanical interaction with correspondingly narrow-spaced objects ., Additionally , the moving cell body described a helical path of similar amplitude ( 3 . 6±0 . 2 µm ) , even without obstacles being present ( Fig . 3B ) , revealing that hydrodynamic drag alone translates into rotation of the entire cell ., In fact , the trypanosomes exploited the mechanical resistance of pillars periodically acting on opposite sides of the cells for efficient locomotion , The rotation of the parasites allowed interaction of the flagellum with the environment equally in all three dimensions , thereby maximizing the probing space of the flagellar tip ( M . Engstler et al . , unpublished ) ., It is the asymmetry of the cell body that causes trypanosomes to rotate ., From its origin at the flagellar pocket to the anterior pole of the cell , the flagellum wraps around the cell body in a turn of approximately 180 degrees ( Fig . 3A ) ( Videos S2 , S3 ) ., This turn is completed after a flagellar course of 11 µm on average ( Video S2 ) ., The remaining process of the flagellum towards the anterior pole of the cell follows the convex side of the cell surface without wrapping around the body significantly ., This means that the helical chirality of the trypanosome is based on a flagellum that makes less than one complete turn around the cell; so the trypanosome ( trypanon gr . \u200a=\u200aauger ) itself does not resemble a corkscrew , but its mode of motion does ( Fig . 3B , C ) ., The rotation becomes evident through high-speed fluorescence microscopy analysis ., Tracking the relative position of the flagellum along the cell body reveals that it always entered the focal plane on the right side of the cell , respective to the direction of movement , and left it on the left side ( Video S4 ) ., Furthermore , by time-dependent tomography , 3D-representations of cells were successfully calculated from high-speed time-lapse series ., This novel approach can produce a valid 3-dimensional model of a given body only if this object rotates unidirectionally with constant angular velocity ., For a trypanosome the angular rotation was determined to be ( 50±10 ) degrees per flagellar beat , proving that the parasites exhibit a rotating type of motion ( Fig . 4 ) ., These results are perfectly consistent with the measured beat frequency of ( 18 . 3±2 . 5 ) Hz ( n\u200a=\u200a60 ) , and a mean rotational frequency of ( 2 . 8±0 . 4 ) Hz ( n\u200a=\u200a58 ) ., This corresponds to ( 8±2 ) single flagellar beats for a full rotation ., Video S5 shows representative high-speed data sets selected from several hundred xyt-image series ., For persistently moving cells , the trypanosome flagellum produced waves that moved unidirectional from the flagellar tip to the base with a constant frequency of about 18 Hz ., Bihelical flagellar waves were not observed 14 ., While the amplitude was increasingly damped by the cell body , a postulated change in the frequency of the flagellar beat 21 could not be confirmed ( Fig . 5 ) ., Due to the virtual absence of inertia at very low Reynolds numbers 22 , 23 , every single flagellar beat produced a distinct and immediate propulsive force ., The resulting locomotion could be visualized in discrete steps , due to the simultaneous rotation of the cell body and the therefore helical path around the axis of movement ( Fig . 5A , Video S8 ) ., The unidirectional rotation of the flagellar beat plane is illustrated in Fig . 5B , in comparison to a theoretical bihelical mode 14 ., Although the overall impression of these models appears rather similar , the physics underlying these two types of motion are fundamentally different ., The major propelling force of trypanosome movement is produced by the beat of the free anterior part of the flagellum ., This force in the anterior direction , together with the hydrodynamic drag force consequently produced in the opposite direction , causes rotation of the posterior flagellar 180° left-hand turn and accordingly of the attached cell body ., In this context it may be worth mentioning that we did not find any support for the existence of an undulating membrane , which is generally thought to drive trypanosome motion ., For various reasons the existence of such a flexible , fin-like extension between cell body and the attached flagellum in T . brucei may be doubted ., Firstly , to the best of our knowledge neither the literature nor our own experiments ( see for example Fig . 3A and Videos S2 , S3 ) provide any compelling microscopic evidence for the presence of an undulating membrane ., Secondly , the physical tethering of the paraflagellar rod to parts of the subpellicular cytoskeleton renders the connection between flagellum and cell body inflexible to the extent of contrasting with the concept of an undulating membrane driving trypanosome motion ., After having detailed the movement of continuously forward swimming cells , we examined trypanosomes not being persistently propelled by tip-to-base beats ., In micropillar arrays , about half of swimming trypanosomes were observed to move backwards by effectively reversing their flagellar beat direction ( Video S11 ) ., The unique switch to base-to-tip beats is characteristic of trypanosomes and has been postulated to accompany tumbling phases and reorientation of the cell body in cell culture 19 , 20 ., Here , for the first time , we show that continuous base-to-tip beats result in persistent backward movement ., The base-to-tip beating trypanosomes were translocated in the posterior direction with each beat and the cells rotated counter-clockwise , viewed in the direction of movement ( Video S11 ) ., This means the flagellar propulsion and the consequent rotation of the cell body observed in forward movement were reversed in base-to-tip beating cells ., The frequency for consecutive base-to-tip beats in backward moving cells ( 13 . 1±0 . 8 Hz ) was lower than for forward swimming parasites ( 18 . 3±2 . 5 Hz ) ., The base-to-tip beats typically produced a more irregular wave pattern with frequently higher amplitudes than tip-to-base beats , enabling the flagellum to fold back against itself , producing a hook-like waveform instead of a regular sine-wave ., Therefore , backwards swimming trypanosomes followed paths of variable amplitudes , dependent on the constraints that the viscous surrounding or the presence of micropillars presented ., Persistent backward motion was only observed with cells swimming in collagen networks , in methylcellulose or between narrow-spaced pillar arrays ( Video S11 ) ., Under conditions offering no confinement and hence no mechanical resistance , continuous base-to-tip beating was not sustained ., Instead , base-to-tip and tip-to-base beats permanently alternated ., This resulted in very short intervals of for- or backward motion interrupted by beat reversal ., While beats of changing direction were initiated , the resulting flagellar waves were simultaneously propagated and thereby generated superimposed waveforms ., Thus , the appearance of a so-called bihelical mode of trypanosome motion 14 in fact reflects transition periods , that do not contribute to , but , quite the contrary , interrupt directional motion ., Importantly , the occurrence of base-to-tip beats was observed to result in rotational movement perpendicular to the anterior-posterior cell axis ., In this way , even very few base-to tip beats interrupting the continuous tip-to-base beating of forward moving cells will alter the trypanosomes swimming direction ( Video S12 ) ., In cell culture , i . e . in the absence of confinement , frequent reversals of the flagellar beat direction caused successive directional changes that led to the characteristic , albeit artificial tumbling movement of trypanosomes ( Video S13 ) ., Many of our conclusions are based on the careful interpretation of high-resolution microscopic data , which is a complex procedure , given that it involves deducing 3-dimensional information from 2-dimensional image series ., Therefore , we corroborated our experimental data with a simulation technique combining the recently developed method of multi-particle collision dynamics to simulate viscous fluid flow 24–28 with a triangulated surface model 29 ., The cell body surface was shaped in such a way , that bending potentials could be applied along the long axis of the body , in order to simulate the stiffness of the trypanosomes cytoskeleton ( Fig . 6A , B ) ., A flagellum was defined and a sine wave running from tip to base applied to it , exactly according to our experimental 3d-microscopy data ( Fig . 6C ) ., The flow field created by the cell propelled the body effectively in the opposite direction to the flagellar wave ., The simulation resulted in the rotation of the whole cell body and in a helical swimming path as observed in our experiments ( Fig . 7 , Video S14 ) ., Furthermore , when a sine wave was applied to the flagellum running in the opposite direction ( base to tip ) , the simulated cell body rotated around axes perpendicular to the anterior-posterior axis ( Video S14 ) , exactly as observed above for trypanosomes experiencing intermittent flagellar beat reversals ., Without mechanical interaction with the surrounding viscous fluid , the model cell body in the numerical simulations neither translocated along nor rotated around the anterior-posterior axis ., This behavior confirms that the course of the flagellum along the cell body as well as the direction of flagellar waves are responsible for the basic steps of trypanosome motility , which are modulated in vivo by the physical micro-environment to produce the apparent complex swimming mode ., Having established that the trypanosomes mode of motility is a consequence of its microenvironment leaves us with the question as to whether it is a genetically fixed trait ., We have shown that inversion of swimming direction and flagellar beat reversal can be triggered by knock-down of a single gene product ( DNAI1 ) ( 5; M . Engstler et al . , unpublished ) ., For efficient interaction with the microenvironment , however , any gene product contributing to cell shape and stiffness should be considered , which renders a precise genetic analysis difficult ., However , there is one last observation that might pave the way for a genetic screen ., High-speed analyses clearly showed that all forward swimming cells displayed exactly the same type of rotational movement , in the absence or presence of obstacles ( Fig . 3B , Videos S4 , S5 , S6 , S7 , S8 , S9 , S10 ) ., The amplitude and frequency of flagellar waves are identical for cells swimming in microstructures , in blood or in cell culture medium ., However , the optimal transduction of force into speed is dependent on a three-dimensional array of obstacles ., This means that the mechanical adaptation to the physical topology of the microenvironment could well be genetically fixed , enabling the cells to move with maximal velocities in their natural habitat ., In fact , for trypanosomes the selective pressure to swim faster than in cell culture medium is evident; it is only at velocities greater than 20 µm s−1 that the hydrodynamic removal of host antibodies becomes effective ., The physical forces acting in and around cells are moving into the focus of attention , not only of biologists , but also of physicists and engineers ., Hydrodynamic flow impacts on cell surfaces even at the scale of molecules 5 ., A mechanism exploiting these forces is critical for the survival of a deadly parasite , the African trypanosome ., The highly motile cells produce a hydrodynamic flow field that drags surface-bound host antibodies against the swimming direction towards the site of localized endocytosis ., This mechanism of antibody removal is only effective if the trypanosomes move directionally with a certain threshold velocity 5 ., This speed is usually not reached in cell culture ., Therefore , we assumed that trypanosome motion in the natural habitat should be faster than under artificial conditions ., So far , the effect of the microenvironment on the motion of trypanosomes , or other cells for that matter , has not been considered adequately ., Moreover , it does not even appear to be clear exactly how trypanosomes swim ., For more than 150 years , the parasites have been observed microscopically 30 , 31 , and their motion became accepted to be corkscrew-like ., Thus , an alternative model of trypanosome swimming came as a surprise , in which the cells are driven by the progression of kinks along the cell body axis 14 ., We have quantified in detail the biomechanics of trypanosome swimming , and could not confirm the proposed rocking-type of motion ., By using high-speed and high-resolution fluorescence microscopy we show that the free segment of the attached flagellum generates a major part of the force required for directional movement ., Since the cell is force and torque free at low Reynolds numbers , the generation of locomotive force by the flagellum is well described by the simple resistive force theory 32 ., The force generated by each beat causes immediate translocation as “inertia plays no role whatsoever” 22 , 23 ., The translocation results in directional laminar flow around the trypanosome cell body ., The attached part of the flagellum largely contributes to the cellular asymmetry and chirality , which causes the moving trypanosome to rotate , as any chiral body will be forced to rotate in a laminar flow field ., As has been shown by Shaevitz et al . for Spiroplasma 33 , a bihelical propulsion mechanism requires a periodic change of cellular chirality 34 ., In other words , for trypanosomes to move in such a way the handedness of the cell body would have to change over time ., This is certainly not the case , as we have analyzed several hundred 3D-fluorescence image data sets without identifying a single trypanosome cell with clockwise chirality ( Fig . 3A ) ., The here proposed complex mechanism of trypanosome movement was confirmed by conducting elaborate computer simulations with size parameters defined by the experiments ., The sinusoidal two-dimensional beat of the flagellum was applied to a trypanosome cell surface model , established on the basis of three-dimensional microscopic data ., The cell model reacted to the flow field created by its motion and moved as observed in the experiments , rotating and describing a helical path ., This analysis underlines the predictive and confirmatory power of computer simulations for analyzing complex biological traits such as cell motility ., The overall picture of a moving trypanosome indeed resembles a corkscrew ( Fig . 3B , C ) , however , the flagellar tip itself produces a two-dimensional beat ( Fig . 5 ) ., We conclude that the term “plane-rotational” would better describe the complex mode of trypanosome locomotion , even though this term may appear paradoxical at first glance ., In order to investigate how the trypanosomes plane-rotational type of motility performs under conditions of crowding and confinement , we trapped the parasites in arrays of micropillars ., The leading flagellum efficiently pulls the trypanosome through the obstacle matrix , the mechanical interaction between cell body and pillars accelerating the cells to significantly higher velocities ., This speed equals that in blood and is sufficient for the hydrodynamic flow-induced antibody clearance in the host ., Thus , trypanosomes exploit friction forces for efficient locomotion ., These forces were surprisingly also seen to act in the reversal of the trypanosomes swimming direction ., Unidirectional tip-to-base beats effectively propel the cell forwards , while interspersed base-to-tip beats lead to a change of swimming direction ., Continued interference of base-to-tip beats causes cells to tumble , but in the presence of mechanical resistance , successive , uninterrupted base-to-tip beating can be sustained , which directly causes the trypanosomes to swim backwards , also reversing their rotational movement ., Thus , the parasite exhibits versatile three-dimensional movement capabilities , directly exploiting the physical nature of its surroundings ., These can easily be envisaged to be of avail for the trypanosomes movement through tight spaces , as the parasite has to be maneuvered through tissue spaces and has to traverse the blood-brain-barrier ., The environmentally induced reversal of swimming direction is a foolproof way of avoiding dead ends ., The fact that motility contributes to survival in the host and hence represents a virulence factor , points to strong selective pressure acting on the blood parasite ., The plane-rotational trypanosome motility can be regarded as a genetically fixed adaptation to the crowded environment in the host ., However , trypanosome motion is not an easily traceable quantitative trait , since not only the flagellum is causative for locomotion but also the shape and architecture of the entire cell ., The half-turn attachment of the flagellum with its unique tip-to-base beat , the elastic paraflagellar rod , as well as the flexible , cage-like cytoskeleton jointly contribute to the complex motion pattern ., Thus , one could argue that evolution has shaped the trypanosome cell for expedient locomotion in blood ., This , however , is not entirely true , since throughout its complex life cycle the parasite experiences dramatically different micro-environmental conditions ., In mammals , trypanosomes not only thrive in blood , but also swim in tissue spaces , in lymph and even in cerebrospinal fluid ., Obviously , the flow regime and the degree of crowding in these areas vary greatly ., Furthermore , the parasites have to leave the mammal with the blood meal of the tsetse fly ., A series of developmental stage transitions takes place in distinct parts of the insect body ., In order to survive the tour de force through the fly , the parasites must adjust their mode of motility ., To study the cooperative action of molecular and environmental cues controlling the motion pattern during trypanosome development seems a rewarding task ., The unusually homogeneous cell surface architecture and the ease of genetic manipulation render African trypanosomes ideal objects for studies at the crossroads of micro-/nanoflow physics , membrane biochemistry and genetics ., Lastly , understanding the microenvironment-dependence of trypanosome motility could offer unforeseen ways of combatting one of the most neglected tropical diseases , the African sleeping sickness 35 ., Wildtype bloodstream form ( BSF ) Trypanosoma brucei brucei , strain 427 36 , Molteno Institute Trypanozoon antigen type 1 . 6 , were cultivated in suspension at 37°C , 5% CO2 in HMI-9 medium , including a final volume of 10% FCS ( Sigma-Aldrich , Taufkirchen , Germany ) ., Cells were kept in the exponential growth phase at a density less than 5×105 cells/ml by dilution with fresh culture medium ., Bovine blood cell | Introduction, Results, Discussion, Materials and Methods | Blood is a remarkable habitat: it is highly viscous , contains a dense packaging of cells and perpetually flows at velocities varying over three orders of magnitude ., Only few pathogens endure the harsh physical conditions within the vertebrate bloodstream and prosper despite being constantly attacked by host antibodies ., African trypanosomes are strictly extracellular blood parasites , which evade the immune response through a system of antigenic variation and incessant motility ., How the flagellates actually swim in blood remains to be elucidated ., Here , we show that the mode and dynamics of trypanosome locomotion are a trait of life within a crowded environment ., Using high-speed fluorescence microscopy and ordered micro-pillar arrays we show that the parasites mode of motility is adapted to the density of cells in blood ., Trypanosomes are pulled forward by the planar beat of the single flagellum ., Hydrodynamic flow across the asymmetrically shaped cell body translates into its rotational movement ., Importantly , the presence of particles with the shape , size and spacing of blood cells is required and sufficient for trypanosomes to reach maximum forward velocity ., If the density of obstacles , however , is further increased to resemble collagen networks or tissue spaces , the parasites reverse their flagellar beat and consequently swim backwards , in this way avoiding getting trapped ., In the absence of obstacles , this flagellar beat reversal occurs randomly resulting in irregular waveforms and apparent cell tumbling ., Thus , the swimming behavior of trypanosomes is a surprising example of micro-adaptation to life at low Reynolds numbers ., For a precise physical interpretation , we compare our high-resolution microscopic data to results from a simulation technique that combines the method of multi-particle collision dynamics with a triangulated surface model ., The simulation produces a rotating cell body and a helical swimming path , providing a functioning simulation method for a microorganism with a complex swimming strategy . | African trypanosomes swim incessantly in the bloodstream of their mammalian host ., We have asked the question how these parasites actually manage to swim and manoeuver in an environment that is so amazingly crowded by blood cells and that reveals rapidly varying fluid flow speeds that are 50–20 . 000 times faster than the trypanosomes swimming speed ., Our experiments suggest an astute mechanism by which trypanosomes have perfectly adapted to their hostile microenvironment ., We found that the pathogens can readily adjust the beating direction of their single flagellum in response to purely mechanical cues ., In the blood they exploit the spacing and shape of blood cells for very efficient forward movement that is required for host antibody clearance ., When the parasites get trapped , i . e . in the extracellular matrix , they reverse the beating direction and consequently move backwards ., The mechanism of flagellar beat switch is unique in nature and represents a genetically fixed trypanosome virulence factor ., By introducing innovative technological advances , we have been able to quantify this complex cell behavior with unprecedented spatial and temporal resolution ., These include the first numerical simulation of a cell of this complexity , extending the protozoans suitability as a model organism for the regulation of flagellar and ciliary motility . | cell motility, parasite evolution, microbiology, host-pathogen interaction, flagellar motility, biomechanics, parasitology, parastic protozoans, cell mechanics, biophysics simulations, zoology, microbial pathogens, biology, biophysics, trypanosoma, biological fluid mechanics, protozoology, molecular cell biology | null |
journal.pntd.0004005 | 2,015 | Clinical Epidemiology of Buruli Ulcer from Benin (2005-2013): Effect of Time-Delay to Diagnosis on Clinical Forms and Severe Phenotypes | Buruli ulcer ( BU ) , caused by Mycobacterium ulcerans , is the third most common mycobacteriosis worldwide , after tuberculosis and leprosy 1 ., BU pathogenesis is mediated by mycolactone , a potent polyketide-derived macrolide that triggers apoptotic cell death 2 and is associated with the necrotic nature of the disease 3 ., BU mostly affects people in tropical countries in Africa 4 , America 5 , Asia 6 and Australia 7 ., Although no official estimate of global incidence is available at present , West Africa is the main endemic area , with 1967 new cases reported by Côte dIvoire , Ghana , and Benin in 20138 ., BU is a devastating necrotising skin infection characterised by pre-ulcerative lesions ( papules , nodules , plaques and edematous infiltration ) , which commonly develop into ulcers with undermined edges and can spread to an entire limb 9 and can also affect the bone ( osteomyelitis ) 10 ., Moreover , within these clinical presentations , more aggressive severe forms of BU , such as multiple lesions , larger lesions or higher World Health Organization ( WHO ) categories have been described 11 , although underreported and less understood ., Epidemiological studies on M . ulcerans transmission , on BU risk factors and on the host immune status , suggest that the variable frequency of BU and its distinct clinical forms are related to:, i ) age;, ii ) gender;, iii ) preferential anatomical site;, iv ) water contact; and, v ) regional occurrences 12 , 13 , 14 , 15 , 16 ., To date , a reduced number of risk factors underlying the severe BU phenotypes had been reported ., HIV co-infection is one of the few examples ., Some studies revealed an increased BU prevalence among HIV patients , especially those presenting large lesions and osteomyelitis 17 , 18 ., Specifically , low CD4 cell counts were significantly associated with larger lesions and patients with a CD4 cell count below 500 cell/mm3 took twice as long to recover from BU when compared with individuals with a normal CD4 cell count 19 ., Other risk factors , such as hypoproteinemia 11 and anemia 20 were also identified to be associated with severe forms of BU disease ., In addition , the delay in seeking medical care and the late medical diagnosis of BU have been proposed to account for the disease presentation 21 , 22 , 23 , 24 ., In fact , in BU endemic regions the culture and beliefs are powerful factors that affect proper medical intervention , as patients preferentially seek treatment from traditional practitioners , or herbalists 22 ., On top of this , the lack of knowledge on the available treatments and their effectiveness , the financial constraints during hospitalisation , fear of treatment , and poor access to health facilities are also important aspects delaying the pursuit of proper treatment 25 , 26 ., Indeed , delay in seeking medical care has been previously associated with the distinct BU clinical forms ., Taking into consideration that the time from progression of a pre-ulcer to an ulcer is variable and can range from a few weeks to several months ( e . g . estimated average time of 30–90 days ) 27 , it was established that individuals with non-ulcerated forms had a median delay of 30 to 45 days , while individuals with ulcers presented a 60-day delay and patients with osteomyelitis up to 90 days 28 ., Thus , the more advanced and destructive ulcerated forms and osteomyelitis are associated with longer delay-periods , while non-ulcerated forms are more common in patients with recent infection 28 , justifying the importance of early diagnosis and treatment for the disease ., Nonetheless , more aggressive , severe clinical presentations of BU , such as large lesions ( >15cm in diameter ) and multifocal lesions , have also been described 11 , although the underlying pathological mechanisms are yet unclear 29 ., While this can be associated with characteristics of the patient itself ( genetic susceptibility/health status ) or with the virulence of the infecting strain , it is also rational to question the influence of the delay in health seeking on the appearance of the more severe forms of BU ., To our knowledge , the latter aspect is yet to be studied ., Therefore , to uncover whether the time-lapse between the first remembered symptoms and clinical diagnosis is associated to disease severity , we retrospectively analysed a cohort of 476 laboratory-confirmed BU treated cases discovered in a highly endemic area in Allada , Benin , between 2005 and 2013 ., Ethical approval ( clearance Nu 018 , 20/OCT/2011 ) for integrating studies on BU was obtained from the National Ethical Review Board of the Ministry of Health in Benin , registered under the Number IRB0006860 ., The Centre de Dépistage et de Traitement de lUlcère de Buruli ( CDTUB ) —Allada and the national BU control program authorities approved access to the registry ., All data analyzed in this study was anonymized ., We retrospectively collected clinical data from 476 laboratory-confirmed BU patients of CDTUB in Allada , Benin—between January 2005 and December 2013 ., At the moment of diagnosis , parameters such as age , gender , major clinical form ( nodule , plaque , edema , ulcer or osteomyelitis ) and multifocal presentations were registered ., For mixed clinical forms , the most severe lesion was considered the major clinical form ., Additionally , lesion size ( cm , considering major diameter ) , WHO category 30 ( Category 1: maximum lesion diameter <5cm; Category 2: maximum lesion diameter 5–15cm; and Category 3: minimum lesion diameter >15cm associated or not with osteomyelitis and/or multifocal lesions and/or at a critical site ) , lesion site ( upper or lower limb , trunk , head and/or neck ) and laboratory confirmation tests ( culture of M . ulcerans from the lesion , histopathology with the presence of acid-fast bacilli , or highly specific IS2404 real-time PCR ) were taken into consideration ., The HIV status was also retained for the present study and excluded from analysis if positive ., Delay in seeking medical care ( time between first symptoms or signs remembered and medical attendance ) was also recorded ., The time of seeking medical care was defined as the moment of diagnosis and treatment initiation ., All included patients completed antibiotherapy according to the WHO recommendations and were treated with surgical procedures 30 ., Explanatory and descriptive analysis of the study cohort was performed based on the following variables: age at the moment of the BU diagnosis; gender; clinical form ( ulcer , plaque , edema , nodule and osteomyelitis ) ; lesion site; and lesion severity ., Severe phenotypes were defined as multifocal lesions ( more than one lesion ) ; large lesions ( diameter >15cm ) or Category 3 lesions ( minimum lesion diameter >15cm associated or not with osteomyelitis , multifocal lesions and/or at a critical site ) as classified by the WHO recommendations ., Median comparisons were performed through one-way ANOVA’s ( Brown-Forsythe and Welch , when applicable ) using age , gender , site of lesion , clinical BU form; and lesion severity as explanatory variables and time-delay seeking medical care ( days , using means and medians distribution in each group ) as a dependent variable ., Unadjusted and adjusted ( for age-cutoff value 15 years of age- and gender binary or linear logistic regression models ) odds ratios were then calculated to explore the effects of time-delay in diagnosis into the clinical form of BU lesions , and particularly into severe phenotypes of BU ., We systematically fit the model , controlling age ( dichotomized or ordinal ) and gender with the considered time-delay ( to seek medical attendance ) as explanatory variables for each of the clinical lesions and severe phenotypes defined for BU ., All the described analyses were obtained using IBM SPSS Statistic v . 22 ., A result was considered significant for p<0 . 05 ., The BU cohort ( CDTUB , Allada , Benin ) , comprising 476 cases , had laboratory BU confirmation by at least one laboratory diagnostic test , as recommended by the WHO ., Results were positive for IS2404 RT-PCR in 430 ( 90 . 3% ) cases and Ziehl-Neelsen staining in 327 ( 68 . 7% ) cases ., All cases were HIV negative ., The median age at diagnosis was 12 years ( IQR: 7–24 years; mean 17 . 9 ± 16 . 3 years ) , with 321 ( 67 . 4% ) patients 15 years old or under ., Although the overall gender ratio of the patients was balanced ( 245 51 . 5% male ) ( Table 1 ) , a major distortion of this ratio was recorded as a function of age , with males being predominant in younger patients and females in older patients ( OR 2 . 99 , 95%CI 2 . 00–4 . 46 , p = 0 . 0001 ) ., Specifically , male patients accounted for 193 ( 60 . 1% ) of the patients younger than 15 , but only 52 ( 33 . 5% ) of those were over 15 ( Table 1 ) ., Considering the dominant clinical BU form per patient , 4 ( 0 . 8% ) presented nodules ( Fig 1A and 1F and Table 1 ) , 24 ( 5 . 0% ) presented edema ( Fig 1B and 1F and Table 1 ) , 125 ( 26 . 3% ) presented plaques ( Fig 1C and 1F and Table 1 ) , and 320 ( 67 . 2% ) presented ulcers ( Fig 1E and 1F and Table 1 ) ., Osteomyelitis was diagnosed in 5 patients ( 1 . 1% ) , and was considered the most relevant form in 3 of the patients ( 0 . 6% ) ( Fig 1E and 1F and Table 1 ) ., Concerning the site of lesions , 256 ( 53 . 8% ) patients presented lesions on the lower limbs , while 171 ( 35 . 9% ) had lesions on the upper limbs ( Fig 2 and Table 1 ) ., Atypical sites ( head , neck and/or trunk ) accounted for 49 ( 10 . 3% ) patients ( Fig 2 and Table 1 ) ., Site of the lesion and relative age , gender and dominant clinical form distribution are represented in Fig 2 ., Regarding the observed severe forms of BU , 22 ( 4 . 6% ) patients presented lesions in more than one localization ( Fig 3A and Table 1 ) , while 142 ( 29 . 8% ) patients presented lesions larger than 15cm in major diameter ( Fig 3B and Table 1 ) ., The WHO category 3 is a broader classification given that it comprises patients with multiple lesions , lesions with a diameter >15cm associated or not with osteomyelitis and/or lesions at a critical site ., Taking into account these criteria , we recorded 315 ( 66 . 2% ) patients in category 1+2 and 161 ( 33 . 8% ) in category 3 ( Fig 3C and Table 1 ) ., The different clinical presentations , as well as the severe forms of these lesions , were subjected to age and gender adjustments ( Tables 2 and 3 , respectively ) ., No significant interference was recorded in the binary logistic regression , except for upper body lesions ( upper limb , head or neck ) , for which there was an overrepresentation of younger ages ( OR 0 . 986 , 95%CI 0 . 974–0 . 998 , p = 0 . 018 ) ( Table 2 ) ., The overall mean time-delay to seek medical care was 101 . 1 days ( 95%CI 86 . 3–117 . 0 ) ( S1 Table ) ., Since the variable time-delay does not follow a normal distribution ( Kurtosis = 48 . 2; Skewness = 6 . 4 ) , median variations were considered to compare the distinct behavior of dependent variables ., Time-delay to seek medical care was indistinct for male and female gender ( p = 0 . 538 ) ( Fig 4A and S1 Table ) : median was 60 days IQR 30–90 for both genders ., However , age was associated with significantly different delay times ( p = 0 . 004 ) ( Fig 4B and S1 Table ) ., Median was 60 days IQR 30–120 for patients over 15 years old at the moment of the diagnosis; while time-delay was 45 days IQR 30–90 for patients with 15 years of age or under ., Time-delay was also related to the clinical form of the disease ( Fig 5 ) ., Median was 32 . 5 days IQR 30–67 . 5 for non-ulcerated forms ( nodule , edema , and plaque ) ; 60 days IQR 20 . 0–120 . 0 for ulcerated forms; and 365 days IQR 228–548 for bone lesions ., When the time-delay among patients with non-ulcerated versus ulcerated forms was compared , we confirmed significant discrepancies ( p = 0 . 009 ) ( Fig 5B and S1 Table ) ., In addition , among the non-ulcerated clinical forms , edema was significantly associated with longer time-delays when compared with others non-ulcerated forms ( median 45 days , IQR 30–105 versus 30 days , IQR 30–60 , respectively , with p = 0 . 03 ) ( S1 Table ) ., Even when age and gender were adjusted in binary logistic regression , we observed an increased risk of developing ulcerative lesions as each day/month passed ( Table 4 ) ., Considering severe forms of BU , none of the aggressive phenotypes were considered related to significantly different delay times to seek medical care: multifocal lesions ( median 90 days , IQR 56 . 3–217 . 5 , p = 0 . 09 ) ( Fig 6A and S2 Table ) , larger lesions with diameter >15cm ( median 60 days , IQR 30–120 , p = 0 . 92 ) ( Fig 6B and S2 Table ) or category 3 WHO classification ( median 60 days , IQR 30–150 , p = 0 . 20 ) ( Fig 6C and S2 Table ) , when compared with unifocal ( median 60 days , IQR: 30–90 ) , small lesions ( diameter ≤15cm ) ( median 60 days , IQR 30–90 ) or WHO category 1+2 lesions ( median 60 days , IQR 30–90 ) , respectively ( Fig 6A–6C and S2 Table ) ., Finally , when systematically fit within binary ( dichotomized variables ) ( Table, 5 ) or linear ( Table, 6 ) logistic regression models controlling for age and gender , time-delay to seek medical care remained statistically insignificant with respect to the occurrence of the most aggressive severe clinical forms ., BU pathogenesis is related with necrosis of the subcutaneous tissue associated with mycolactone , the potent cytotoxic/immunosuppressive toxin produced by M . ulcerans 3 ., Initial pre-ulcerative lesions ( papules , nodules , plaques and edematous infiltration ) can evolve into ulcers and progressively spread over significant extensions of the body 9 or even affect the bone 10 ., Large national studies in West African countries , namely Ghana 31 , Benin 28 , 29 , 32 and Côte D`Ivoire 33 , included the largest BU cohorts studied thus far and provided information about the age and gender of patients , site of lesions and the major clinical forms—providing further clues on the evolution of BU pathology ., The majority of these studies used distinct methodologies ( retrospective and/or prospective cohorts; cross-sectional ) and a descriptive approach , with a large proportion of diagnoses being retrospective and scar-based ., Here , we strictly consider laboratory-confirmed BU patients ., Concerning the BU clinical forms ( papules , nodules , plaques , edematous infiltration , ulcers and osteomyelitis ) , the observations of our study globally fit the variances reported in those larger cohorts ., Specifically , we confirm that BU is mainly a paediatric disease ( median age of diagnosis 12 years with IQR: 7–24 years and mean of 19 . 7 years ) ; with a predominance of lesions on the lower limbs ( 53 . 8% ) ; a predominance of ulcerative forms ( 67 . 2% ) ; and with an equilibrium between genders ., In addition , there is a distinct distribution of gender when age is considered , with males being overrepresented in younger patients , reproducing data from previous studies 15 , 29 ., Osteomyelitis and edematous forms are classified as belonging to the spectrum of BU presentations , although some authors consider them to be more severe clinical forms 32 , 34 , 35 , 36 ., Regarding osteomyelitis , a great variance in prevalence is described and further complexity is added when suspected non-confirmed cases of bone involvement are included in the analysis ., Indeed , reported prevalence values of bone disease related to BU were as high as 29 . 5% 37 and 36 . 1% 38 ., However , when only confirmed osteomyelitis cases were considered , prevalence decreased with values ranging between 6% 29 and 20% 39 in Africa and only 1% in Australia 40 ., Moreover , HIV infection seems to favour the occurrence of osteomyelitis 17 ., In the present study , osteomyelitis lesions only occurred in 1 . 1% of the at-risk population , which could be related to the fulfillment of confirmed diagnosis criteria ( e . g . x-ray or surgical evidence ) and the absence of the HIV co-infection selection criteria ., Eedematous lesions manifest as diffuse , extensive , usually non-pitting swelling with ill-defined margins involving part or all of a limb or other part of the body 41 ., Cases of edematous M . ulcerans infection can be misdiagnosed as bacterial cellulitis leading to delays in diagnosis , progression of disease , increased morbidity and increased complexity and cost of treatment ., Additionally , edema is often self-perceived as not being a relevant health problem , therefore delaying seeking medical attention ., In previous studies , prevalence was determined to be between 2 . 5% 42 and 12 . 5% 31 ., In our study , edematous forms accounted for 5% of the studied population , fitting with the prevalence reported in similar cohorts 31 , 34 , 35 , 36 , 42 ., Within to the above described clinical BU presentations , more aggressive , severe clinical presentations have been described 11 , although the underlying pathological mechanisms are yet unclear 29 ., In our study , within the severe phenotypes , 33 . 8% of the patients were in WHO category 3; 4 . 5% presented multifocal forms; and 29 . 8% of the patients presented lesions >15cm in major diameter ., Regarding multifocal lesions , previous studies describe highly variable prevalences ( e . g . 2 . 0% -11 . 1% ) 40 , 42 , 43 , 44 , 45 , 46 ., Moreover , in our African cohort , we verify that age does not associate with multifocal lesions , conversely to an Australian cohort 40 ., Regarding lesion size , only a few studies report large lesions as a specified studied variable , since these lesions are usually included in category 3 lesions ., However , when considered separately , their prevalence ranged between 11 . 1% 47 and 36 . 0% 29 , while category 3 lesions have been reported to range between 19 . 7% 48 and 60 . 0% 39–values replicated in the present study ., The effect of time-delay in seeking medical care for BU patients is a relevant issue for public health and patient management ., Our observations in a cohort of laboratory-confirmed cases of BU show that gender was not related with distinct behavior in seeking specific medical care and that younger patients , mainly through their parents/legal tutors , spent less time seeking medical attention prior to diagnosis ( median 45 versus 60 days , for the group ≤15 years old versus >15 years old respectively , p = 0 . 004 ) ., In line with previous African studies , we found that more advanced ulcerative forms were related to the delay in seeking medical care ., Remarkably , and contrary to what one would expect , we found that multifocal lesions , larger lesions or WHO category 3 lesions may be considered distinct clinical entities since the time-delay in seeking medical attention had no significant role in disease progression ., As a matter of fact , in Africa , time-delay was seen as a marker of accessibility to medical care and , in fact , some studies compare time-lapse before and after interventional politics on health care improvement ., In West Africa , studies reported a time-delay between 42 38 and 84 days 44 , 49 , taking into consideration all clinical forms ., Specifically , a Beninese study reported distinct clinical forms relating to time-lapse since first symptoms were remembered 28 ., Time-delay was shorter for non-ulcerated clinical forms ( median 30 to 46 days ) , than for ulcerated forms ( median 61 days ) and larger for osteomyelitis ( median 91 days ) ., In Australian studies , time-lapse until medical care was reported to be much shorter—between 14 days ( IQR 0–6 weeks ) 40 and 42 days ( ranging from 2 and 270 days ) 50 ., In this distinct health-care reality , determinants for delay in seeking medical care were related to atypical sites of lesions , associated with an increased complexity in medical BU diagnosis ., Interestingly , in Australian patients , ulcerated versus non-ulcerated clinical forms did not experience significantly different time lapses ., Moreover , independently of the advances in diagnosis and clinical management , there was no variation in time-delay between 1998–2004 and 2005–2011 ., In Southern America , the time-delay reported among Peruvian BU patients was between 1 and 8 months 51 ., Overall , our observations in a cohort of laboratory-confirmed cases of BU , strengthening previous observations and show that the time-delay in seeking medical care is related to the more advanced ulcerative forms , further justifying early diagnosis and treatment ., Notably , we additionally show that time-delay was not significantly associated with more severe phenotypes of BU , such as multifocal lesions , larger lesions or WHO category 3 lesions ., Indeed , our results demonstrate that after initial progression lesions become stable regarding size and focal/multifocal progression ., Therefore , in future studies on BU epidemiology , severe clinical forms should be systematically considered as distinct phenotypes of the same disease and therefore subjected to specific risk factor investigation ., These results further highlight that intrinsic regulatory mechanisms , such as the host immune response and local biochemical and physical factors , most likely have relevant roles in determining severe phenotypes , justifying more structural immune-related and bacterial genetic studies . | Introduction, Materials and Methods, Results, Discussion | Buruli Ulcer ( BU ) is a neglected infectious disease caused by Mycobacterium ulcerans that is responsible for severe necrotizing cutaneous lesions that may be associated with bone involvement ., Clinical presentations of BU lesions are classically classified as papules , nodules , plaques and edematous infiltration , ulcer or osteomyelitis ., Within these different clinical forms , lesions can be further classified as severe forms based on focality ( multiple lesions ) , lesions’ size ( >15cm diameter ) or WHO Category ( WHO Category 3 lesions ) ., There are studies reporting an association between delay in seeking medical care and the development of ulcerative forms of BU or osteomyelitis , but the effect of time-delay on the emergence of lesions classified as severe has not been addressed ., To address both issues , and in a cohort of laboratory-confirmed BU cases , 476 patients from a medical center in Allada , Benin , were studied ., In this laboratory-confirmed cohort , we validated previous observations , demonstrating that time-delay is statistically related to the clinical form of BU ., Indeed , for non-ulcerated forms ( nodule , edema , and plaque ) the median time-delay was 32 . 5 days ( IQR 30 . 0–67 . 5 ) , while for ulcerated forms it was 60 days ( IQR 20 . 0–120 . 0 ) ( p = 0 . 009 ) , and for bone lesions , 365 days ( IQR 228 . 0–548 . 0 ) ., On the other hand , we show here that time-delay is not associated with the more severe phenotypes of BU , such as multi-focal lesions ( median 90 days; IQR 56–217 . 5; p = 0 . 09 ) , larger lesions ( diameter >15cm ) ( median 60 days; IQR 30–120; p = 0 . 92 ) or category 3 WHO classification ( median 60 days; IQR 30–150; p = 0 . 20 ) , when compared with unifocal ( median 60 days; IQR 30–90 ) , small lesions ( diameter ≤15cm ) ( median 60 days; IQR 30–90 ) , or WHO category 1+2 lesions ( median 60 days; IQR 30–90 ) , respectively ., Our results demonstrate that after an initial period of progression towards ulceration or bone involvement , BU lesions become stable regarding size and focal/multi-focal progression ., Therefore , in future studies on BU epidemiology , severe clinical forms should be systematically considered as distinct phenotypes of the same disease and thus subjected to specific risk factor investigation . | Buruli Ulcer ( BU ) is a neglected disease caused by Mycobacterium ulcerans ., Clinical presentations of BU lesions are classically classified as papules , nodules , plaques and edematous infiltration , ulcer or osteomyelitis ., Within these different clinical forms , lesions can be further classified as severe forms based on focality ( multiple lesions ) , lesions’ size ( >15cm diameter ) or WHO Category ( WHO Category 3 lesions ) ., There are studies reporting an association between delay in seeking medical care and the development of ulcerative forms of BU or osteomyelitis , but the effect of time-delay on the emergence of lesions classified as severe has not been addressed ., To address both issues , and in a cohort of laboratory-confirmed BU cases , 476 patients from a medical center in Allada , Benin , were studied ., In our cohort , we validated previous observations , demonstrating that time-delay is statistically related to the clinical form of BU , namely ulcers and osteomyelitis ., However , time-delay is not related with more severe phenotypes , implying that severe clinical forms of BU should be considered as distinct phenotypes of the same disease and subjected to specific risk factor investigation . | null | null |
journal.pcbi.1004300 | 2,015 | Design of Protein Multi-specificity Using an Independent Sequence Search Reduces the Barrier to Low Energy Sequences | In contrast to SSD , multistate design ( MSD ) minimizes the free energy of multiple protein conformations ( “states” ) simultaneously ., This enables negative design , which involves destabilizing a certain conformation to shift relative occupancy to alternate conformations , which is useful in designing proteins with binding selectivity ., MSD has been applied successfully in a number of cases , including the design of a protein conformational switch 9 , design of selective b-ZIP binding peptides 10 , and design of an enzyme with DNA cleavage specificity 11 , among others 12 , 13 ., All MSD algorithms have at their core a fitness function that defines the favorability of a given sequence based on its corresponding energy in each state ., The major challenge in fixed backbone MSD is efficient optimization of side chain rotational isomer ( “rotamer” ) placement , using the fitness function as the objective function ., As more states are considered it becomes increasingly difficult to find the minimum energy sequence on a fixed backbone ., As the same sequence on all states is constrained , extensive sampling in sequence and rotamer space is required ., This is often accomplished via thorough but slow genetic algorithms 12 , 14 , 15 ., This difficulty in reaching the global minimum in a basic fixed backbone design problem precludes the possibility of using alternate sampling strategies , such as iterating between backbone minimization and rotamer optimization ., However , these techniques have been used in SSD to great effect and are often critical to find the lowest energy conformation and sequence 1 , 8 ., In result , MSD algorithms can arrive at an incorrect solution even after successful sequence optimization just because the fixed backbone precludes the lowest energy sequence and conformation from being sampled ., This issue can be partially resolved by the inclusion of multiple backbone conformations as separate states 16 ., However , there is a need for a method that can more efficiently reach the optimal MSD solution for an arbitrary number of input states without relying on the commonly held “fixed backbone assumption” ., To this end , we have developed a novel MSD algorithm , referred to as REstrained CONvergence in multi-specificity design ( RECON ) ., The algorithm is based on a different conception of MSD , wherein each state independently explores sequence space to reach its energetic minimum ., A step-wise increasing convergence restraint is applied such that corresponding positions in different states converge on the same amino acid ., By encouraging sequence convergence between different states rather than enforcing a single sequence , we hypothesize that energetic barriers to the fittest solution collapse , reducing the ruggedness of the energetic landscape in a MSD problem to SSD-like complexity ., In result the search efficiency and speed are substantially increased allowing for the sampling of additional degrees of freedom ., Further , we hypothesize that including backbone conformational sampling reduces the chance that the low energy and possibly correct solutions are excluded from the search space ., The RECON algorithm allows separate states to explore their own local sequence and conformational space to optimize free energy , while restraining corresponding residues in different states with a convergence restraint to encourage sequence convergence ., Convergence restraints are kept small in early rounds , to allow each state to explore its own lowest energy sequence , and ramped up in later rounds to encourage sequence convergence between different states ., This is followed by a greedy selection step , which evaluates all candidate amino acids at positions that fail to converge , and selects the one that results in the lowest fitness when applied over all states ., This greedy selection is included in order to ensure that one multi-specific sequence is generated from each design trajectory ., Backbone minimization steps can be included between design rounds to relieve slight clashes between side chains ., Pseudocode describing the implementation of the algorithm is shown in Fig 1 ., Individual states optimize rotamer placement using a simulated annealing Monte Carlo search , sampling from a predefined rotamer library 17 , 18 ., However , we emphasize that this method can be applied to any multi-specificity problem using an arbitrary optimization method and scoring function ., By allowing each state to determine its optimal sequence independently , we can collapse the energy barrier to reaching a “compromised” sequence that results in low energy in all states ., We propose a scenario in which encouraging sequencing convergence in this way can reduce the energetic barrier and enable convergence on a low energy solution ( Fig 2 ) ., In this scenario , two separate mutations from residue identity A to B are needed for the lowest fitness over both states ., Each single mutation will encounter a high energy penalty and rarely selected by a genetic algorithm–only when both mutations are stochastically placed together will the solution emerge , which may take a large number of evaluations ., However , when sequence convergence is encouraged rather than enforced , each state will identify an intermediate solution in early rounds , and in later rounds the most favorable solution will be selected from the differing states ., This collapses the barrier on the pathway to a favorable solution and reduces the steps necessary to find that solution ., To benchmark RECON , we considered two types of design problems ., In the first , mature antibodies derived from a common germline gene were entered into MSD in complex with their target antigens ., It has been shown that MSD of mature antibodies results in a higher rate of germline sequence reversion than SSD , implying that the germline sequence is near-optimal for polyspecificity 19 , 20 ., Therefore , we designed each antibody against its respective targets and used germline sequence recovery as an indirect measure of the rate of recovery of an optimal solution ., We used antibodies derived from three different germline genes—VH1-69 , VH3-23 , and VH5-51 ., The number of antibody-antigen complexes per germline gene ranged from 3 to 6 ( Table 1 ) ., The second task was to design a set of “promiscuous” proteins , proteins that have been crystallized in complex with multiple binding partners , against each of these partners ., Similar to polyspecific germline antibodies , promiscuous proteins have been shown to have a native sequence that is near-optimal for binding to all of the partners 14 , 22 ., Therefore an effective MSD protocol would result in a high rate of native sequence recovery ., A set of five promiscuous proteins derived from a study done by Humphris et al . was used 14 , in addition to two broadly neutralizing anti-influenza hemagglutinin antibodies ( Table, 2 ) 23 , 24 ., Benchmark cases were designed using three separate design methods ., First , design was performed using RECON with a fixed backbone ., Fixed backbone design has to this point been the standard in MSD due to the complexity involved in recalculating rotamer interactions for each backbone movement ., However , using fixed backbone design alone is prone to false negatives , as sequences that may be highly favorable with a small shift in backbone conformation are discarded ., One of the unique advantages of RECON is its ability to incorporate iterative rounds of rotamer packing and backbone minimization ., Therefore , we included such an iterative protocol as the second approach in our benchmark ., For comparison purposes , all complexes were also designed using the existing MSD application in Rosetta ( MPI_MSD ) , which operates on a fixed backbone 15 ., MPI_MSD differs from RECON in that it uses a genetic algorithm to create and advance mutations and a user-defined fitness function to assess fitness of each sequence ., However , as both methods are built into the Rosetta framework , they sample from the same rotamer library and use the same scoring function and are therefore suitable for comparison ., In addition to native sequence recovery , we used the fitness of the top ten designs , defined as the sum of Rosetta energies of all complexes , to analyze how effectively each protocol reached an energetic minimum ., This fitness function has been previously used in studies of design of protein multi-specificity 14 ., We use the term “design” to refer to sequence optimization of existing protein-protein complexes—however , it is important to note that these sequences were not experimentally characterized , and results reported are purely in silico ., For common germline gene-derived antibodies , RECON was consistently able to recover the germline sequence at a higher rate than MPI_MSD ( Table 3 and Fig 3 ) ., Germline sequence recovery for RECON ranged from 55–94% using fixed backbone and 51–95% using backbone minimization , while recovery for designs using MPI_MSD ranged from 32–64% ., When comparing RECON fixed backbone to MPI_MSD , it appeared that designs created by RECON , although higher in native sequence content , were also energetically less favorable ., We therefore subjected all fixed backbone designs to a single round of Rosetta relax energy minimization to relieve frustrations and allow for direct comparison of fitness of RECON incorporating backbone minimization to fixed backbone designs ( S1 Table ) ., These post-minimization fitness values show that the energetic gap between RECON- and MPI_MSD-generated designs was substantially closed , and that designs generated by any method occupied similar ranges of fitness ., We observed that MPI_MSD tended to produce designs with the lowest fitness—however , it is important to note that rotamer optimization within Rosetta is a stochastic process , with no guarantee of reaching the global minimum ., Therefore a protocol that performs hundreds of rounds of rotamer optimization , such as MPI_MSD , would be expected to produce better energies than one performing four rounds of optimization , such as RECON , independent of the sequence identity of structures being optimized ., RECON was able to recover the native sequence at a very high level for all promiscuous protein complexes–native sequence recovery ranged from 72–100% and 73–100% , using fixed backbone and backbone minimization , respectively ., MPI_MSD generated designs with native sequence recovery ranging from 70–94% ( Table 4 and Fig 3 ) ., In most cases fitness of designs generated by RECON fixed backbone and MPI_MSD were very similar , suggesting that both methods have reached the energetic minimum ( S1 Table ) ., Even though all methods reached a similar level of native sequence recovery and energetic fitness in a majority of these benchmark cases , RECON was able to reach these minima by searching a compressed sequence space , allowing for increased computational efficiency ., We hypothesized that gradually ramping the convergence restraints will allow for sequence divergence in early rounds of design and enforce convergence in later rounds , leading to an improved result as it smoothens the energy landscape ., To confirm the effects of gradually increasing the weight of the convergence restraint , we performed a control in which sequences were designed independently for each state with no convergence restraint , followed by sequence selection by the greedy selection used at the end of RECON ( S2 Table ) ., This greedy selection algorithm performed significantly worse than RECON with gradual ramping convergence restraints , with worse native sequence recovery in all benchmark cases but one ., In addition , in many benchmark cases fitness was significantly worsened for designs generated by this greedy selection protocol ., These results indicate that ramping convergence restraints throughout the design protocol is critical for the increased performance of RECON ., Based on the decreased performance of this greedy selection algorithm , it would be expected that RECON works best at positions where amino acids converge between different states by the end of the protocol and are not greedily selected ., We therefore evaluated the convergence of amino acids at each position for the VH5-51 benchmark set ., We report the number of times a position failed to converge in 100 design trajectories for the 30 designed positions in this benchmark set ( S3 Table ) ., The results suggest that most positions tend to be consistent in their patterns of convergence , and that the majority ( 21 out of 30 ) reach a common amino acid solution by the end of the protocol ., The results of the greedy selection protocol suggest that failure to converge leads to a decrease in performance of the algorithm and selection of non-native amino acids ., We therefore compared germline sequence recovery for positions that failed to converge in at least half of the design trajectories , as compared to those that converged in more than half of the trajectories , to determine whether these positions are substantially decreasing overall germline sequence recovery ( S3 Table ) ., Surprisingly , positions that failed to converge actually showed a higher rate of germline sequence recovery than those that were able to converge through the application of convergence restraints ( S3 Table ) ., These results indicate that , although the greedy selection algorithm should not be applied without first ramping convergence restraints to encourage convergence , the use of greedy selection for positions that fail to converge is not a limiting factor for obtaining high native sequence recovery ., In the scenario proposed in Fig 2 , we hypothesize that RECON is able to circumvent high-energy intermediate sequences by encouraging rather than enforcing sequence convergence ., We therefore analyzed the sequence trajectory of an example from the FI6v3 benchmark to support this scenario ( Fig 4A ) ., In early rounds , the two states diverge in sequence to explore their own energy landscapes ., As restraints are increased in later rounds the two states converge on a compromised sequence that is the multi-specific solution for both , only adopting mutations when they are beneficial to both states ., Although fitness values continue to decrease after encountering the compromised sequence , this is primarily due to the stochastic nature of rotamer optimization , such that increased optimization will result in a lower score ., We focused on a set of complementary mutations that diverged in early rounds with a low convergence restraint , to test the hypothesis that the sequence preference of one state results in a high energy on the other state , and vice versa ( Fig 4A , highlighted in red ) ., We found that the sequences preferred by state 1 ( TSY ) and state 2 ( QQW ) indeed resulted in higher energy when forcing one state to adopt both sequences than when each state was allowed to adopt its own sequence ( Fig 4B ) ., This lowers the barrier to reaching the “compromised” sequence , adopting residues favorable to both state 1 and state 2 , which in this case is the sequence QQY ., Although this barrier is not as large as proposed in Fig 2 , we expect that this barrier will be lower in cases where two binding partners have highly similar binding surfaces , as is the case in our benchmark sets ., However , when binding surfaces are more dissimilar , and therefore finding compromise residues is more critical to a favorable binding energy , we expect this barrier to be larger , and the benefit of an independent sequence search to be even greater ., In addition to measuring the sequence recovery and energetic fitness , we compared the computational efficiency of these three design protocols ., We argue that , although in certain cases all methods were able to reach the same energetic minimum , RECON provides an added benefit in that it reached this minimum in a fraction of the time required to run MPI_MSD ., To this end we compared CPU hours of runtime for generating the previously discussed designs ( Table 5 ) ., As expected , RECON using a fixed backbone was the most efficient of the three protocols , followed by RECON incorporating backbone minimization , and MPI_MSD ., This increase in efficiency is due to the reduction in search space by allowing each state to adopt its own sequence ., We hypothesize that RECON is able to operate at higher efficiency by restricting sampled sequences to more relevant sequence space ., We further believe that our conception of “relevant” sequence space is reflected in an ensemble of biologically observed sequences , and that RECON should recover not only a native protein sequence , but also biologically tolerated mutations ., To address this question we generated a position-specific scoring matrix ( PSSM ) of amino acid frequencies in evolutionarily related proteins to each benchmark protein using a PSI-Blast query 25 ., Among the promiscuous proteins we restricted this analysis to non-antibodies , since the full-length sequence of a mature antibody is unlikely to have a large number of meaningful evolutionary counterparts ., However , since antibodies in the common germline-encoded benchmark set were only designed in positions deriving from the VH gene , we were able to derive a PSSM from other common VH-encoded antibodies in the database ., We then compared the PSSM to the amino acid frequency in corresponding positions in designed sequences to estimate how well the design protocol mimicked evolution ., We measured agreement of sequence profiles using a modified Sandelin-Wasserman similarity to yield a percent similarity for each designed position that could then be averaged over the protein 26 ., Fig 5A shows a comparison of positions in the VH5-51 benchmark where designs either agreed or disagreed with evolutionary sequence profiles—the degree of agreement could then be quantitated by the percent similarity calculated over each position ., We found that RECON was able to create sequences that more closely mirrored natural sequence variation than MPI_MSD ( Table 6 and Fig 5B ) ., Averaging over the benchmark cases , we observed 69 , 73 , and 57% similarity to evolutionary sequence profiles using RECON fixed backbone , backbone minimized , and MPI_MSD , respectively ., This pattern was especially strong in benchmark cases with large numbers of designable residues , as the number of designed residues correlated positively with the improvement of RECON over MPI_MSD in recapitulating evolutionary sequence profiles ( Fig 5C ) ., When comparing the four largest benchmark cases by number of designable residues ( three common germline-derived antibodies and the PAPD complex ) , RECON shows a marked improvement over MPI_MSD in recovery of evolutionary sequence profile ( Fig 5D ) ., Although this result is not significant due to a small sample size , it is suggestive of the additional benefit provided by RECON when applied to large , computationally intensive design problems ., We hypothesize that this is due to compressed sequence space explored by RECON ., When design problems are relatively small , the genetic algorithm employed by MPI_MSD is able to efficiently search through sequence space for a low-energy solution ., However , when the sequence space increases in a large design problem the compressed sequence search is more advantageous ., We have shown that designs generated by RECON tend to more closely represent the evolutionary sequence profiles of our benchmark proteins when compared with MPI_MSD ., We propose that this is accomplished via a more focused sequence search within the biologically relevant space ., To further support this claim , we have analyzed the sequence space searched by RECON and MPI_MSD and compared it to the final output sequences of the top ten designs for the VH5-51 benchmark set ( S1 Fig ) ., We generated the sequence space profile by including any residue that was sampled at any step of the design protocol at each position , and then compared this profile to the final sequences among the top ten designs ., Presumably the most efficient algorithm would only sample the sequences that are eventually selected as low energy solutions , resulting in a similarity of 100% between sequence space explored and output designed sequences ., Therefore we used this similarity as an indicator of the degree of “wasted” sequence space , which is explored but never part of a low energy solution ., Comparison of the profiles generated by RECON on a fixed backbone and MPI_MSD show that RECON explores space much more closely constrained to the final low energy sequences , with a similarity score of 92% , as compared to 80% for MPI_MSD ., This further supports the claim that RECON searches a compressed search space to encounter a low energy multi-specific solution ., The algorithms RECON and MPI_MSD feature substantial differences in sequence and structure at many positions of the output design models , particularly in the common germline antibody benchmark sets ., We hypothesized that this difference in preference may be due to a failure by MPI_MSD to exhaustively search through sequence space in a large design problem ., Concurrently we expect that the sequences selected for by RECON are actually lower in overall fitness ., We present structural analysis of three positions , residues 32 , 33 , and 74 in the VH3-23 benchmark , to support this claim ., Position 32 showed a preference for tyrosine in RECON-generated designs , whereas MPI_MSD prefers glycine ., Tyrosine is able to fill a cross-interface gap in the 1S78 complex , and can establish hydrogen bonding to an amide nitrogen across the interface ( Fig 6A ) ., This additional hydrogen bonding produces a large drop in fitness for this residue across all states ( -1 . 85 versus -5 . 97 REU ) ., Interestingly , tyrosine is the germline residue at this position , and was only recovered using RECON with backbone minimization—both RECON fixed backbone and MPI_MSD favor glycine at this position ., Position 33 also showed difference preferences between design methods—alanine was favored by MPI_MSD , whereas RECON favored serine ., Serine results in a lower overall fitness due to additional hydrogen bonding with a glutamine residue on the heavy chain CDR3 loop of the antibody ( Fig 6B ) ., At this position , alanine is the germline residue—however the per-residue fitness values indicate that serine is able to stabilize this loop in the 3BN9 complex without compromising stability of the other states ( Fig 6B , fitness shown in parenthesis ) ., Lastly , position 74 showed a preference for threonine in RECON-generated designs , as opposed to serine in MPI_MSD-generated designs ., Threonine is able to establish cross-interface hydrogen bonding in the 1S78 complex without causing clashes in other states , whereas serine is somewhat surprisingly not positioned to create this interaction ( Fig 6C ) ., This is partially due to backbone movements in the RECON-generated structure that position the hydroxyl group for optimal hydrogen bond geometry ., In addition to hydrogen bonding , threonine scores more favorably on the basis of increased van der Waals attractive forces of the additional methyl group with surrounding atoms ., At this position , asparagine is the germline amino acid , which was recovered by neither RECON nor MPI_MSD ., From our initial benchmark results , we did not observe a difference in evolutionary sequence similarity for designs created with fixed backbone versus backbone minimization protocols ( Fig 5D ) ., However , as previous reports have shown the utility of incorporating backbone motion into a design protocol 8 , 27–29 , we hypothesized that the initial minimization of structures before entering them into multi-specificity design reduced the impact of alternating backbone minimization with design ., We hypothesize that backbone movement should have a larger impact on design of structures that have not been pre-minimized ., To test this hypothesis , we repeated the benchmark with structures that had not been pre-minimized , and performed multi-specificity design with three protocols:, 1 ) fixed backbone design ,, 2 ) alternating design with minimization of φ , ψ , and χ angles , and, 3 ) alternating design with backrub movements ., The backrub motion involves rotation of a rigid backbone around axes between nearby Cα atoms , and has been shown to recapitulate alternative backbone conformations in high-resolution crystal structures 30 as well as improving prediction of the conformation of point mutant side chains 31 ., We predicted that a design protocol including backrub motions between design rounds should result in the highest agreement to evolutionary sequence profiles , given the sampling of more biologically relevant conformational space than simple minimization ., We therefore analyzed the similarity to evolutionary sequence profiles for the top ten designs produced by the three methods and compared to evaluate whether backbone motion in this context confers any additional benefit ., As expected , incorporating backrub movements results in a statistically significant increase in similarity to evolutionary profiles as compared to a fixed backbone protocol or one involving minimization ( Fig 7 ) ., This agrees with previous studies indicated that backrub motions are able to sample biologically relevant conformational space , and shows that backrub motions can be incorporated in a multi-specificity context to provide more robust results in terms of evolutionary sequence recovery ., In previous works involving both germline antibodies and promiscuous proteins , the difference in sequence recovery between sequences generated by single state and multi-specificity design has been analyzed 14 , 19 ., Multi-specificity design in both cases was shown to recover the native or germline sequence at a higher rate than single state design , supporting the proposition that the increased performance of multistate design justifies the increased computational complexity ., Given the increased performance of RECON in native sequence recovery , we hypothesized that multi-specificity design performed by RECON would result in a larger difference in germline vs . mature sequence recovery in the common germline antibody dataset ., We therefore performed fixed backbone single state design for each complex in this dataset and calculated recovery of the germline sequence and the mature antibody sequences ., We can recover the difference in germline and mature sequence recovery as observed in 19 , and show that design performed by RECON results in a larger difference between germline and mature sequence recovery compared to MPI_MSD ( S2 Fig ) ., We can therefore conclude that in these cases RECON is more robust at generating germline-like , multi-specific sequences compared to MPI_MSD ., We have developed and benchmarked a new method for multi-specificity design , REstrained CONvergence in multi-specificity design ( RECON ) ., This algorithm operates by allowing each state to search sequence space independently with a restraint system that gradually encourages convergence between different states on a common sequence ., Allowing each state to adopt a unique sequence reduces the space of sequences required to search in order to find a native-like low energy solution ., In two separate benchmark sets consisting of ten total cases , we were able to show that RECON , both with and without iterative backbone minimization cycles , was able to more accurately recapitulate the native , multi-specific sequence of input proteins than the existing MSD application in Rosetta , MPI_MSD ., In addition , we analyzed agreement of designed sequences with observed evolutionary sequence profiles to measure how well MSD simulates natural sequence tolerance ., In large design problems with many residues being optimized simultaneously , RECON was able to create sequences that more closely mirrored the natural distribution of sequences seen in evolutionary profiles ., In this study we analyzed the degree of convergence of a designed protein sequence profiles with the natural sequence variation seen in evolutionary homologs ., It is well known that many proteins tolerate a wider variety of sequences than simply the native sequence 13 , 29 , 32 , 33—therefore a major goal of multi-specificity design is to recover not only the native sequence of a protein , but also sequence variations that are tolerated by all binding partners ., We found that RECON is able to recover evolutionary sequence profiles more effectively in large complexes—however , it is clear from analysis of sequences sampled by each method ( S1 Fig ) that MPI_MSD is exploring a much larger sequence space ., In certain cases this diversity of sampling may be desired , especially in cases where the interface with both binding partners is compatible with a large number of sequence polymorphisms ., Our benchmark cases suggest that sampling near the energy minimum for each individual state is sufficient to recover the sequences compatible with all states ., However , in cases where generating sequence diversity is at a premium , for example to explore the tolerated sequence space of a given backbone , it may be advantageous to use RECON and MPI_MSD as complementary approaches ., RECON in the current study was used with the standard Rosetta simulated annealing rotamer optimization protocol 34–however , other rotamer optimization methods have shown superior performance in certain instances ., For example , MPI_MSD uses a modified form of the FASTER algorithm 35 , referred to as backbone-minimum-energy conformation followed by single-residue perturbation/relaxation ( BMEC-sPR ) 15 ., Leaver-Fay et al . compared the effectiveness of these two algorithms and found that BMEC-sPR consistently reached the global minimum solution in a higher proportion of cases 15 ., Additional rotamer optimization algorithms have been adapted for use in MSD , such as dead-end elimination 36 , probabilistic graphical models 37 , and iterative batch relaxation/single perturbation and relaxation 35 ., The benefit of RECON is that it can be adapted to work with any single state-compatible rotamer optimization method , as communication between different states is conducted solely by the restraint system ., This opens up the possibility of adapting many more optimization methods for MSD ., One important benefit of RECON is the ability to incorporate backbone motion into an MSD protocol ., Traditionally protein flexibility in MSD has been modeled by including multiple backbone conformations as input states 9 , 13 , 16 , 32 , 38 ., This is a reasonable strategy for running MSD using RECON ., However , RECON offers the benefit that each state can be subject to additional backbone minimization between design rounds ., When incorporating backbone motion into design the conformational and sequence space explodes , making it difficult to reach a global minimum ., However the fact that RECON reduces the sampling needed to reach the optimal sequence allows for more search space to be explored ., We have shown that incorporating backbone flexibility in the form of backrub motions can improve accuracy of sequences when applied to un-minimized structures ., Single state design protocols have successfully incorporated backbone movement , allowing the introduction of mutations that would have been unfavorable on the original backbone 1 , 8 , 27 ., The ideal protocol for flexible backbone design remains elusive , considering the different methods of backbone perturbation 28 , 30 ., In addition it remains unclear how to best alternate fixed backbone sequence optimization with backbone motion 39 ., RECON opens up the possibility of incorporating these backbone design methods into an MSD context ., One of the most challenging aspects of MSD is the inclusion of unfavorable states to destabilize ., The current implementation of RECON is limited in scope compared to approaches such as MPI_MSD due to the inability to perform negative design to disfavor | Introduction, Results, Discussion, Methods | Computational protein design has found great success in engineering proteins for thermodynamic stability , binding specificity , or enzymatic activity in a ‘single state’ design ( SSD ) paradigm ., Multi-specificity design ( MSD ) , on the other hand , involves considering the stability of multiple protein states simultaneously ., We have developed a novel MSD algorithm , which we refer to as REstrained CONvergence in multi-specificity design ( RECON ) ., The algorithm allows each state to adopt its own sequence throughout the design process rather than enforcing a single sequence on all states ., Convergence to a single sequence is encouraged through an incrementally increasing convergence restraint for corresponding positions ., Compared to MSD algorithms that enforce ( constrain ) an identical sequence on all states the energy landscape is simplified , which accelerates the search drastically ., As a result , RECON can readily be used in simulations with a flexible protein backbone ., We have benchmarked RECON on two design tasks ., First , we designed antibodies derived from a common germline gene against their diverse targets to assess recovery of the germline , polyspecific sequence ., Second , we design “promiscuous” , polyspecific proteins against all binding partners and measure recovery of the native sequence ., We show that RECON is able to efficiently recover native-like , biologically relevant sequences in this diverse set of protein complexes . | The ability to design a new protein with a desired activity has been a longstanding goal of computational biologists , to create proteins with new binding activity or increased stability ., An even more ambitious goal is multi-specificity design , which extends general protein design by creating a sequence that has low energy with multiple binding partners ., We have developed a new algorithm for multi-specificity design that more efficiently finds a low energy sequence for all complexes ., This increased efficiency enables simulation of biologically relevant motion between binding partners , such as backbone movement and shifts in orientation ., We show that our algorithm outperforms existing approaches , and compare the predicted low energy sequences to the sequences naturally seen through evolution of each protein ., We find that this algorithm is able to more accurately represent the scope of sequences that are found in biological contexts ., This method can be applied to design new proteins with the ability to bind multiple distinct partners . | null | null |
journal.pntd.0005999 | 2,017 | Integration of enteric fever surveillance into the WHO-coordinated Invasive Bacterial-Vaccine Preventable Diseases (IB-VPD) platform: A low cost approach to track an increasingly important disease | Enteric fever ( typhoid/paratyphoid ) is a major cause of mortality and morbidity in many low and middle-income countries ., In 2015 , it was estimated to cause about 178 , 000 deaths and 17 million illnesses; 85% of all cases occur in three countries—Bangladesh , India and Pakistan 1–3 ., The primary causative agents of enteric fever are Salmonella enterica serovars Typhi ( typhoid ) and Paratyphi A , B , and C ( paratyphoid ) 4 ., Enteric fever was one of the largest killers during the pre-antibiotic era , but case fatality rates have decreased from 30% to less than 1% with the use of effective antibiotics 4 ., In recent years , however , there has been an increasing number of reports on the rise of antimicrobial resistance of Salmonella Typhi/Parathyphi 5 ., If this trend is not interrupted , untreatable infections with case fatality rates much higher than those experienced in the last few decades are likely to occur ., This evolving crisis calls for urgent guidelines for institution of effective treatment and prevention policies; however , a scarcity of accurate data on burden and epidemiological and antimicrobial resistance trends impedes evidence-based policy decisions ., For example , despite the availability of the typhoid Vi polysaccharaide vaccines ( ViPS ) that provide protection in >2 years children in Bangladesh , where the main etiology of blood stream infections is Salmonella Typhi/Paratyphi A 6–8 , they are not used programmatically ., Due to poor reception by the public and clinicians , two pharmaceutical manufacturers ultimately removed their products from market; consumption in 2013 was only about 2 , 500 doses ( personal communication ) ., Patients and physicians largely rely on empirical antibiotic therapy ., This lack of data primarily stems from a lack of robust surveillance for enteric fever in endemic areas , which are mainly in resource-poor countries that cannot afford to establish and/or sustain new surveillance programs for this disease 9 ., Credible disease estimates can help policy-makers and government officials to prioritize interventions ., Unlike for enteric fever , most countries already have , or are in the process of introducing , pneumococcal conjugate vaccine ( PCV ) and Haemophilus influenzae type b ( Hib ) vaccine to combat invasive diseases caused by Streptococcus pneumoniae and Hib , respectively ., The decisions to implement these vaccines have been facilitated by surveillance data generated by the Gavi’s PneumoADIP and Hib Initiative and the World Health Organization ( WHO ) -coordinated Global Invasive Bacterial-Vaccine Preventable Diseases ( IB-VPD ) Surveillance Network ., The IB-VPD surveillance assesses the burden of pneumonia , meningitis and sepsis in children ≤59 months ., This surveillance system captures data on S . pneumoniae , Hib and Neisseria meningitidis ( from blood and cerebrospinal fluid , CSF ) and monitors long-term trends , including the impact of vaccines 10 ., As of 2016 , the system is in place in 56 countries 11 ., Notably , enteric fever is not part of this surveillance platform ., Since 2009 , Bangladesh has been operating four high-performing WHO IB-VPD sentinel surveillance sites in Dhaka ( two sites ) , Chittagong and Mirzapur ., Epidemiological data generated from these sites facilitated introductions of Hib and pneumococcal vaccines in Bangladesh 11 ., In this study , we aimed to evaluate the logistics , cost and sustainability of leveraging the ongoing WHO-coordinated IB-VPD platform for enteric fever surveillance by broadening the inclusion criteria of the original surveillance ., The study was conducted in two urban hospitals , Dhaka Shishu ( Children ) Hospital ( DSH ) and Shishu Shasthya ( Child Health ) Foundation Hospital ( SSFH ) , which are sentinel sites of the WHO IB-VPD platform in Bangladesh ., These are the two largest pediatric hospitals in the country ., DSH provides primary to tertiary care to patients aged 0–18 years; it acts both as a primary point of care and the major referral hospital of the country ., SSFH acts as the primary point of care for children aged 0–14 years ., Together , they have 840 beds , 246 ( 29% ) of which are reserved for families who are unable to pay ., The hospitals are located close to each other and primarily cater to the same catchment area of Mirpur , Dhaka ., In Bangladesh , using the WHO-coordinated IB-VPD surveillance system , we monitor pneumonia , meningitis and sepsis in children under 5 years old admitted in the in-patient departments ( IPD ) using protocols described elsewhere 12 ., In brief , IPD patients are assessed by research physicians and are considered eligible for WHO IB-VPD surveillance study if they meet clinical definitions for meningitis , pneumonia or sepsis ( WHO’s IB-VPD inclusion criteria , Table 1 ) 12 ., Blood specimens are collected from eligible cases at the discretion of the attending physicians ., Only cases from which a specimen is collected are enrolled ., In January 2012 , we expanded the existing surveillance platform to include surveillance of enteric fever; based on WHO case definition , we set inclusion criteria for enrollment as fever of ≥102°F for ≥3 days with no clinical manifestations of pneumonia , meningitis or sepsis ( Table 1 ) 13 ., Here we report our findings from this expanded surveillance between January 2012 and December 2016 ., To test the success rate of capturing enteric fever cases using the expanded IB-VPD platform , we additionally enrolled all cases ( of any age and any clinical presentation ) from the IPD whose blood culture yielded growth of Salmonella Typhi/Paratyphi A in the laboratories but were not enrolled in the active hospital surveillance systems ., Microbiology laboratories of the two hospitals are also the diagnostic laboratories that provide service to all patients ., They receive and culture all blood specimens referred by physicians in the hospitals ., Blood culture results of all cases were communicated to the study physicians by the laboratory staff as soon as they were available ., Study physicians at DSH and SSFH reviewed these cases of Salmonella Typhi/Paratyphi A that were not captured through the IB-VPD or enteric fever surveillance systems and enrolled them in the study ., Clinical information of these culture-confirmed enteric fever cases were collected through the study physicians’ assessment within 24 hours of laboratory confirmation from hospital charts 12 ., Blood cultures were performed using standard methods as described earlier 14 ., In brief , 2–3 ml blood was aseptically obtained and inoculated into trypticase soy broth supplemented with sodium polyethanol sulphonate ( 0 . 25% ) and isovitalex ( 1% ) ., All blood collections were performed by trained hospital phlebotomists and stringent measures were taken to avoid contamination by skin flora during phlebotomy ., Blood collection was only performed once per patient ., Incubated blood culture bottles were sub-cultured on the 2nd , 3rd and 5th days of incubation ., Identification of Salmonella Typhi/Paratyphi A isolates was confirmed after standard biochemical tests and agglutination with Salmonella species and serovar specific antisera ( Ramel , Thermo Fisher Scientific , USA ) ., Antibiotic susceptibility tests were conducted and interpreted according to most the recent Clinical and Laboratory Standards Institute ( CLSI ) guidelines for each antibiotic ., Disc diffusion methods were used for ampicillin , cotrimoxazole , chloramphenicol , ciprofloxacin , azithromycin and ceftriaxone ., In addition , microbroth dilution method was performed for ciprofloxacin ., All data were entered into EpiData ( The EpiData Association , Odense , Denmark ) and analyzed using Stata 13 ( STATACorp , College Station , TX ) ., Data were analyzed to examine the distribution of culture-positive etiological agents of meningitis , sepsis , pneumonia and enteric fever cases , and their clinical and epidemiological features ., Age distribution and antimicrobial susceptibility of enteric fever cases were also examined ., No statistical tests were performed in this study ., The cost of integrating enteric fever surveillance into the ongoing IB-VPD surveillance was calculated based on the wet-laboratory resources required to process 1 , 699 additional blood specimens and the additional personnel appointed ., While calculating wet laboratory expenses , extra items required for culture and identification of Salmonella Typhi and Salmonella Paratyphi A using biochemical and serological tests with specific antisera were considered ., Two research physicians ( 50% time each , one at each facility ) and two research assistants ( 50% time each , one at each facility ) were needed to aid in assessment of the additional cases that became eligible and were enrolled due to the additional enteric fever surveillance inclusion criteria ., The protocols were approved by the ethics review committees of the Bangladesh Institute of Child Health , Dhaka Shishu ( Children ) Hospital , Bangladesh ., Blood samples were collected as part of routine clinical care and written consent was obtained from parents or caregivers of all participants for other aspects of the study , including collection of data and use of specimens for additional laboratory analysis ., Based on the original inclusion criteria defined by the WHO-coordinated IB-VPD surveillance , 10 , 130 cases were identified with possible invasive bacterial diseases ( sepsis , pneumonia , meningitis ) ( Fig 1 ) ., Amongst them , 5 , 185 ( 52% ) cases were enrolled on collection of blood specimen for culture; 171 ( 3 . 3% , 171/5 , 185 ) cases were culture-positive ( Fig 1 ) ., Of the organisms detected , 55% ( 94/171 ) were either Salmonella Typhi ( 50%; 85/171 ) or Salmonella Paratyphi A ( 5 . 2%; 9/171 ) ., The other predominant organisms were S . pneumoniae ( 29%; 49/171 ) and N . meningitidis ( 1 . 2%; 2/171 ) ( Fig 2 ) ., A full list of organisms detected in this original platform are listed in S1 Table ., Using the added inclusion criteria of ≥102°F for ≥3 days for enteric fever , a further 2 , 744 cases with suspect enteric fever were identified , who did not meet the clinical criteria of WHO-defined meningitis , sepsis or pneumonia ( IB-VPD surveillance ) ., Blood specimens were collected from 1 , 699 ( 62% , 1699/2744 ) of them and a total of 358 ( 21% , 358/1699 ) specimens were culture-positive ., Amongst the culture-positive cases , Salmonella Typhi was the predominant etiology ( 85% , 305/358 ) , followed by Salmonella Paratyphi A ( 12% , 44/358 ) ( Fig 1 ) ., Thus , the additional case definition increased the number of all blood cultures performed by 25% {1699/ ( 1699+5185 ) }and this resulted in a five-fold increase in the detection of enteric fever cases ( from 94 to 443 in five years ) ., Of the remaining nine culture-positive specimens captured with the additional inclusion criteria , six were cases of S . pneumoniae infections , two of non-typhoidal Salmonella infections and one of Escherichia coli infection ( Fig 2 , S1 Table ) ., In all age groups seeking care at the in-patient departments of the hospitals , a total of 311 culture-confirmed enteric fever cases ( 263 Salmonella Typhi and 48 Salmonella Paratyphi A ) cases were identified in the laboratories that were not enrolled through the WHO IB-VPD surveillance or the added enteric fever surveillance ., Amongst them , 9% ( 28/311 ) were from children 2–59 months old , the WHO IB-VPD age group ., The combination of the IB-VPD and the added fever ≥ 102°C for ≥ 3 days in the case definition captured 443 cases in this age group ., Thus , the proposed expanded surveillance , using the IB-VPD platform , captured 94% ( 443/471 ) of the blood culture-confirmed enteric fever cases among 2–59 months old in-patients ., The remaining 91% ( 283/311 ) culture-confirmed enteric fever cases obtained in the laboratories were from children > 59 m ., In total , 754 culture-confirmed typhoid and paratyphoid cases were enrolled from the in-patient departments in this study ., Of them , 283 ( 38% ) cases were children aged > 59 m and hence only captured after obtaining laboratory results ., Out of 754 total culture-positive typhoid/paratyphoid cases identified , 471 enteric fever infections were in the target IB-VPD population of 2–59 m old children ., In this age group , 36% ( 170/471 ) of the infections occurred in the first 24 months of life ., Children in their second year of life seemed most vulnerable with 27% ( 125/471 ) cases occurring in children aged 12–23 months ( Fig 3 ) ., Antibiotic susceptibility was tested for all Salmonella Typhi ( N = 653 ) and Salmonella Paratyphi A isolates ( N = 101 ) ., In total , 166 ( 25% ) Salmonella Typhi isolates were multi-drug resistant ( MDR , resistant to chloramphenicol , ampicillin and trimethoprim ) ., No MDR Salmonella Paratyphi A strain was identified ., While 578 ( 89% ) of Salmonella Typhi and 70 ( 69% ) of Salmonella Paratyphi A isolates were susceptible to azithromycin , all strains were susceptible to ceftriaxone ., By contrast , 646 ( 99% ) Salmonella Typhi and 99 ( 99% ) Salmonella Paratyphi A isolates were non-susceptible to ciprofloxacin ., Clinical manifestations of all 754 laboratory-confirmed typhoid and paratyphoid cases were assessed ( Table 2 ) ., Of the 94 cases captured in the original WHO IB-VPD surveillance , 78% ( 73/94 ) had fever of ≥102°F for ≥3 days ., Median duration of fever was five days ., Other manifestations included vomiting ( 30% , 28/94 ) , convulsions ( 37% , 35/94 ) and inability to feed ( 17% , 16/94 ) ., In the 349 cases captured with the added inclusion criteria for enteric fever , where only children with ≥ 102°C fever for ≥ 3 days were enrolled , the median duration of fever was six days and vomiting ( 37% , 130/349 ) was often observed ., Overall , clinical manifestation of enteric fever cases identified by the IB-VPD surveillance appeared more severe than those identified based on expanded fever ≥ 102°C fever for ≥ 3 days criteria , most notably with convulsions ( 37% vs 4% ) and inability to feed ( 17% vs 0% ) occurring at a higher rate ., Amongst the 28 2–59 m children enrolled after laboratory confirmation , who were neither captured by IB-VPD or enteric fever surveillance , physicians observed a fever of ≥102°F in only one case and the duration of fever was <3 days ., Vomiting was the most common manifestation ( 32% , 9/28 ) ., In the >59 months-old group ( N = 283 ) , fever of ≥102°F was observed in 249 ( 88% ) cases and the median duration of fever was six days ., The most common accompanied clinical sign was also vomiting ( 41% , 117/283 ) ., The additional cost for running the enteric fever surveillance leveraging the WHO’s IB-VPD platform was USD 19 , 374 per year for research physicians and assistants , in addition to the base expenditure of USD 167 , 765 of the platform ., The wet laboratory work required additional reagents and resources of USD 25 , 600 per year ., The total of USD 44 , 974 for added enteric fever surveillance is a 27% increase to the annual IB-VPD cost ., In South Asia , Salmonella Typhi/Paratyphi A comprise three-fourths of all isolates obtained from blood cultures of sick pediatric and general populations 6–8 ., Enteric fever can be prevented by improving water , sanitation and hygiene and with effective vaccines ., However , as it is a disease that disproportionately affects resource-poor communities , unless the vaccines are provided through the government health services , the new generation typhoid vaccine carries a risk of not protecting the most vulnerable children in the poorest countries ., Moreover , it remains difficult for policy makers to make evidence-based decisions as historically only a few , small , sporadic studies have addressed disease burden ., Recently , with investments like the Coalition against Typhoid by the Bill and Melinda Gates Foundation , there has been renewed interest among the global community , including industries , in this disease ., The Sabin Vaccine Institute , with support from the Bill and Melinda Gates Foundation , established a hospital-based enteric fever surveillance network in Asia called the Surveillance for Enteric Fever in Asia Project ( SEAP ) , to enable systemic collection of data and fill knowledge gaps on impact of severe enteric fever ., Another surveillance program , Typhoid Fever Surveillance in Africa Program ( TSAP ) , was completed in Africa in 2014 and its follow-on study , Severe Typhoid in Africa ( SETA ) that measures the severity and burden of enteric fever , is underway 15 ., However , such comprehensive population-based surveillance systems are expensive and bear the risk of unsustainability ., For example , the large multi-country study , Pneumococcal Vaccine Accelerated Development and Introduction Plan ( PneumoADIP ) , successfully demonstrated the burden of pneumococcal disease in developing countries and facilitated introduction of the appropriate vaccines 16 ., Nevertheless , due to the high costs , this comprehensive surveillance system was not sustained ., To sustainably monitor trends of enteric fever and the characteristics of its etiologies , sustainable and cost-effective surveillance systems are desirable ., In Bangladesh , we have been able to establish a cost-effective integration of enteric fever surveillance within the ongoing WHO-coordinated IB-VPD surveillance system to generate data on another vaccine preventable disease ., During 2012–2016 , in the IB-VPD surveillance , 55% etiologies of all blood culture positive cases of suspect pneumonia , sepsis and meningitis were Salmonella Typhi and Paratyphi A , despite the fact that IB-VPD does not aim to monitor enteric fever ., Over the same time period , our attempt to identify additional enteric fever cases using added inclusion criteria of fever >102°F for ≥ 3 days increased capture of culture-confirmed enteric fever episodes in 2–59 m children from 19/year to 89/year ., We report a total of 471 cases in this age group , where 94 were captured through the original IB-VPD surveillance and 358 through the added enteric fever surveillance ., An additional 28 cases were found through analysis of laboratory results , which were missed by the inclusion criteria of the surveillance study ., Overall , this integrated IB-VPD and enteric fever surveillances was able to capture 94% ( 443/471 ) of culture-confirmed enteric fever cases in 2–59 m children ., This integration was sustainably managed for five years with no major hurdles and required an increase of 27% in cost; co-sharing of resources and personnel make the proposed surveillance a cost-effective approach ., The data generated from this multi-layered and multi-year study corroborate with results from previous typhoid specific studies from the region ., Previous studies in Bangladesh showed that more than 50% of typhoid cases occur in children under the age of 5 years , similar to the findings of this study 7 , 17 , 18 ., The rate of multidrug resistance in recent years has been reported to be around 20% for Salmonella Typhi strains in Bangladesh with more than 90% non-susceptibility to ciprofloxacin 6 , 19 ., Similar rates were observed in our surveillance ., The surveillance system proposed here is not without limitations ., Firstly , blood cultures were not performed for all suspect cases ., This limitation is intrinsic to all IB-VPD sentinel site based surveillance systems since obtaining blood cultures is mainly dependent on the discretion of the treating physicians ., To improve the blood-culture practice , our study team leveraged state-of-the-art laboratories and provided the test free of cost and the hospital authorities encouraged the treating physicians to advise blood culture ., However , the large number of culture-positive cases ( 471 cases in 2–59 m patients and 283 cases in >59 m patients ) that were detected yielded large amount of information and indicate that the proposed integrated surveillance system can be used to generate high quality epidemiological data and monitor antimicrobial resistance trends ., Secondly , previous studies performed in Bangladesh and other endemic countries of the region show that majority of enteric fever cases seek care at the out-patient departments of hospitals , where they are commonly treated using empirical therapy; as the proposed enteric fever surveillance rides on the ongoing IB-VPD surveillance , only in-patient cases can be captured ., Despite missing a large proportion of cases , such a surveillance program will capture the most severe cases with higher likelihood of hospitalization ., Moreover , the antimicrobial resistance trends learnt from documented cases in this surveillance can guide empirical treatment policies in out-patient departments ., Thirdly , because the surveillance reporting is not population-based , it does not have a denominator and thus does not allow for incidence calculation ., This can be overcome if the data can be linked to a denominator using the low-cost hybrid approach proposed by Luby et al 9 ., This approach combines the existing laboratory diagnosis data conducted in healthcare centers with those from community-based surveillance of utilization of healthcare facilities ( study hospitals ) to generate incidence estimates ., We are currently initiating such a combined surveillance to calculate incidence and generate data relevant to policy decisions ., Additional resources are also being invested to follow-up patients to characterize outcome and estimate disease severity and case fatality rates ., Furthermore , previous work on impact of pneumococcal conjugate vaccine has shown that with such large number of cases in sentinel sites , it is also possible to monitor vaccine impact in hospital-based surveillance systems 20 ., Fourthly , the stated costs for the proposed integration were estimated based upon sites that have considerable experience in bacterial disease surveillance ., It is unclear how much this experience and these costs would transfer to other sites ., Overall , this study demonstrates that enteric fever surveillance can be sustainably and cost-effectively integrated into the original IB-VPD surveillance and the proposed integrated platform will fulfill the objectives of WHO for other invasive bacterial vaccine preventable diseases:, ( i ) to collect data to describe epidemiology and estimate burden ,, ( ii ) to establish a surveillance platform in order to establish baseline rates of disease to measure impact after introduction of vaccines and, ( iii ) to detect and characterize circulation bacterial types 10 ., Establishment of a new and stable surveillance platform is time consuming and expensive and there is a high possibility of failure ., With typhoid conjugate vaccines in sight we recommend that WHO considers the integration of enteric fever surveillance into the present IB-VPD platform ., With such a system in place , areas with known high burden of enteric fever will benefit from better understanding of epidemiological characteristics of the disease and antimicrobial resistance trends for optimal empirical treatment ., We also recommend this integration in areas where the burden of enteric fever is unknown , as this can act as a rapid and cost-effective way to monitor hospitalized children without major changes in infrastructure and or loss of resources if the disease is found to be absent ., Data generated from such surveillance systems will help countries make evidence-based decisions on introduction of upcoming vaccines and prepare for evaluation of vaccine impact studies ., By including enteric fever surveillance into the WHO-coordinated IB-VPD surveillance , robust data can be obtained about the true burden of one of the leading vaccine preventable diseases . | Introduction, Methods, Results, Discussion | Lack of surveillance systems and accurate data impede evidence-based decisions on treatment and prevention of enteric fever , caused by Salmonella Typhi/Paratyphi ., The WHO coordinates a global Invasive Bacterial–Vaccine Preventable Diseases ( IB-VPD ) surveillance network but does not monitor enteric fever ., We evaluated the feasibility and sustainability of integrating enteric fever surveillance into the ongoing IB-VPD platform ., The IB-VPD surveillance system uses WHO definitions to enroll 2–59 month children hospitalized with possible pneumonia , sepsis or meningitis ., We expanded this surveillance system to additionally capture suspect enteric fever cases during 2012–2016 , in two WHO sentinel hospitals of Bangladesh , by adding inclusion criteria of fever ≥102°F for ≥3 days , irrespective of other manifestations ., Culture-positive enteric fever cases from in-patient departments ( IPD ) detected in the hospital laboratories but missed by the expanded surveillance , were also enrolled to assess completion ., Costs for this integration were calculated for the additional personnel and resources required ., In the IB-VPD surveillance , 5 , 185 cases were enrolled; 3% ( N = 171/5185 ) were positive for microbiological growth , of which 55% ( 94/171 ) were culture-confirmed cases of enteric fever ( 85 Typhi and 9 Paratyphi A ) ., The added inclusion criteria for enteric fever enrolled an additional 1 , 699 cases; 22% ( 358/1699 ) were positive , of which 85% ( 349/358 ) were enteric fever cases ( 305 Typhi and 44 Paratyphi A ) ., Laboratory surveillance of in-patients of all ages enrolled 311 additional enteric fever cases ( 263 Typhi and 48 Paratyphi A ) ; 9% ( 28/311 ) were 2–59 m and 91% ( 283/311 ) >59 m ., Altogether , 754 ( 94+349+311 ) culture-confirmed enteric fever cases were found , of which 471 were 2–59 m ., Of these 471 cases , 94% ( 443/471 ) were identified through the hospital surveillances and 6% ( 28/471 ) through laboratory results ., Twenty-three percent ( 170/754 ) of all cases were children <2 years ., Additional cost for the integration was USD 44 , 974/year , a 27% increase to the IB-VPD annual expenditure ., In a setting where enteric disease is a substantial public health problem , we could integrate enteric fever surveillance into the standard IB-VPD surveillance platform at a modest cost . | Typhoid/paratyphoid fever imposes a major global burden , specifically in low-and-middle-income countries ( LMICs ) ., However , it is challenging to implement evidence-based decisions for treatment and prevention because of lack of data from comprehensive surveillance systems , which are often expensive and difficult to sustain ., The WHO has established a global surveillance program called “Invasive Bacterial–Vaccine Preventable Diseases ( IB-VPD ) Surveillance” to capture sepsis , meningitis and pneumonia in under-five children in many LMICs ., Data generated by this program have facilitated introduction of live-saving vaccines and development of treatment strategies ., However , the program does not include typhoid/paratyphoid surveillance ., We tested the feasibility and sustainability of integrating typhoid/paratyphoid surveillance into this program in two leading children’s hospitals in Bangladesh ., By monitoring all patients with signs of typhoid/paratyphoid , we captured 471 laboratory-confirmed episodes in under-five children between Jan 2012 and Dec 2016 ., Blood culture results from all in-patients revealed that the proposed expanded surveillance captures 94% of hospitalized typhoid/paratyphoid cases ., Thirty-six percent ( 170/471 ) of 2–59 m cases were in children <2 years ., Overall , age distribution and antibiotic resistance patterns were consistent with data generated from larger , expensive and typhoid-specific surveillance programs in the region , adding credence to the proposed integration ., Adding typhoid/paratyphoid surveillance to an established invasive disease surveillance platform took advantage of existing infrastructure and resources and as such was easy and cost-effective to implement ., We recommend that WHO considers similar integration in other countries; data generated from such surveillances will help countries make evidence-based decisions on introduction of upcoming vaccines and prepare for impact studies . | inflammatory diseases, medicine and health sciences, body fluids, pathology and laboratory medicine, pathogens, microbiology, salmonella typhi, pulmonology, vaccines, pneumonia, bacterial diseases, signs and symptoms, enterobacteriaceae, infectious disease control, bacteria, bacterial pathogens, infectious diseases, infectious diseases of the nervous system, medical microbiology, epidemiology, microbial pathogens, salmonella, infectious disease surveillance, diagnostic medicine, blood, anatomy, meningitis, fevers, physiology, neurology, disease surveillance, biology and life sciences, organisms | null |
journal.pcbi.1002179 | 2,011 | Computational Modeling of Allosteric Communication Reveals Organizing Principles of Mutation-Induced Signaling in ABL and EGFR Kinases | The phenomenon of allosteric communication is fundamental to many biological processes and is recognized as an important factor governing molecular regulation of signal transduction networks 1 , 2 ., Theoretical and computational studies of allostery in biomolecular systems have witnessed a recent renaissance fueled by the growing interest in the development of quantitative models of allosteric communication in proteins and biological networks 3–7 ., Sequence-based approaches have unveiled that protein allostery may be mediated by coupled motions of evolutionary conserved yet sparse networks of functional residues which constitute signal communication pathways in proteins 8 , 9 ., Recent network-based structural studies have also demonstrated that allosteric pathways may be formed through interactions of evolutionary conserved residues that are energetically coupled to mediate long-range communication 10–12 ., Mechanistic understanding of collective protein motions and allosteric transitions at the molecular level has been significantly advanced by the employment of elastic network models ( ENM ) and normal mode analysis ( NMA ) 13–25 ., These approaches have been further integrated with the information-based Markovian theory of signal propagation 26 , 27 and have provided a generalized formalism of allosteric communication in proteins 28–31 ., Structure-based ENM approaches combined with sequence-based bioinformatics analyses have identified that conserved low-frequency modes of collective motions are robust to sequence variations and capable of transmitting molecular signals over long distances 23–25 , 31 ., Allosteric communication mechanisms can range from a sequential model , where binding of a molecule at one site causes a sequential propagation of conformational changes across the protein , to a fully cooperative model , where structural changes are tightly coupled ., More recently , an intermediate , “block-based” model was proposed , where sparse clusters of closely interacting residues can maintain a weak association to other blocks of residues and thus pass information between more distance regions of a protein 32 ., Collectively , these studies have shown that allosteric networks orchestrating cooperative protein motions can be formed by evolutionary conserved and sparsely connected groups of residues , suggesting that rapid transmission of allosteric signals through a small network of distantly connected residue clusters may be a universal requirement encoded across protein families ., Statistical analyses of motions in allosteric proteins with known inactive and active crystal structures have quantified the magnitude of allosteric effects , revealing a strong preference toward weakly constrained regions such as loops and protein surface regions 33 ., Subsequent graph-based analysis and a global communication network model have shown that small-world allosteric networks have sparse connectivity and long-range protein communication is determined by specific residue clusters playing critical roles in the transmission of functional signals 34 , 35 ., The global communication network ( GCN ) model has integrated tertiary ( residue-scale ) and quaternary ( subunit-scale ) structural changes , providing a more general representation of allosteric communication mechanisms that allowed to simplify atomistic simulations and proven useful in guiding experiments probing allosteric function 35 ., Data mining and machine learning methods using support-vector models have helped to infer rules that can distinguish structural hotspots of functionally important allosteric residues 36 ., Computational biophysics studies of allosteric regulation have explored a functional linkage between simulations of protein dynamics and energetics of allosteric coupling 37–45 ., Thermodynamics-based approaches that linked structural perturbations with free energy changes of allosteric coupling have provided quantitative insights into allosteric mechanisms of conformational switching 46–49 ., A physics-based perturbation method , the Rotamerically Induced Perturbation ( RIP ) , can generate concerted protein motions by applying local torsional perturbations to individual residues 50 ., Nonequilibrium methods can monitor concerted protein motions and determine the distribution of signaling pathways , while avoiding long simulation times required in conventional molecular dynamics ( MD ) simulations 51 ., Graph-based analysis of protein allosteric communication can reduce complexity and yield a convenient characterization of the protein architectures as one-dimensional maps comprised of nodes ( residues ) connected by edges ( inter-residue “interactions” ) 52–61 ., These methods have shown that protein structural graphs form small world networks 52–55 , characterized by high local residue connectivity and a small number of long-range connectivity ., Network and graph-based approaches have been employed in predicting protein-protein interactions 54 , 55 catalytic sites in enzymes 56 , 57 , protein structure , energetics and evolution 58–61 ., The allosteric regulation of protein kinases serves as an efficient strategy for molecular communication and event coupling in signal transduction networks ., The regulatory interactions have a major role in determining the conformational dynamics of the kinase domain and activation mechanisms 62–69 ., Protein kinase regulation may be controlled by a dynamic coupling of two spatially distributed yet conserved and functionally important intermolecular networks between the N-lobe and the C-lobe forming a hydrophobic regulatory spine and a catalytic spine 66–69 ., The wealth of structure-functional studies about protein kinases has demonstrated that protein kinase activity can be tightly regulated via dynamic interconversion between closely related active and highly specific inactive kinase states - a structural hallmark of the kinase domain which is critical for its normal function 70–79 ., High-resolution nuclear magnetic resonance ( NMR ) spectroscopy can complement X-ray crystallography studies by probing protein dynamics on multiple time scales and detecting a site-specific ligand signature that allows differentiation between competitive and allosteric inhibitor binding 80 ., NMR studies have detected protein kinase motions in the active and inactive forms on multiple time scales , suggesting that conformational mobility is vital for regulatory control of kinase activity 81 ., The dependence of chronic myeloid leukemia ( CML ) on the translocated BCR-ABL kinase is associated with unique drug responses to small molecule inhibitors 82 ., The mechanism of protein kinase regulation via dynamic equilibrium between structurally different functional states has been successfully exploited in the discovery of selective inhibitors targeting inactive conformations of the ABL kinase 83–94 ., A large number of point mutations that impair the binding of Imatinib ( Gleevec ) to ABL have been described 83 , 84 , suggesting that some drug resistant mutations could exist before treatment , and may contribute to tumorigenesis ., Structurally conserved gate-keeper mutation ABL-T315I is a dominant cancer-causing alteration , leading to the most severe Imatinib resistance by favoring the active form of the ABL kinase ., These findings guided the design of the second-generation ABL inhibitors Dasatinib and Nilotinib 85–91 ., While these inhibitors are effective against most ABL mutants , the ABL-T315I mutation is still resistant to all three therapies ., Most recently , a third-generation of rationally designed analogs and hybrids of Imatinib and Dasatinib , including Ponatinib , DCC-2036 and HG-7-85-01 92–94 were shown to recognize a broad spectrum of inactive kinase conformations and retained potency against ABL-T315I ., Nevertheless , activating mutations that destabilize the inactive conformation of ABL ( most notably ABL-T315I ) still result in reduced binding affinity of these inhibitors ., Although the vast majority of protein kinase inhibitors bind to the ATP binding site of the catalytic domain , a considerable effort has been recently invested to discover inhibitors associated with a specific kinase and disease 95–100 ., Unlike ATP-competitive kinase inhibitors , allosteric inhibitors typically bind outside the catalytic domain and affect kinase activity by eliciting global conformational transformations , which may confer a greater specificity and allow for a subtle modulation of kinase regulation 101 ., Allosteric regulation mechanisms in protein kinases may include stabilization of the inactive MEK kinases by targeting the adjacent to the ATP binding pocket in MEK-1 , MEK-2 102 , 103 and JNK kinase 104–106; allosteric binding to the myristoyl-binding pocket of ABL and regulation via formation of multidomain ABL-SH2-SH3 complexes 107–114; activation mechanism via formation of regulatory complexes in cyclin-dependent kinase 2 ( CDK2 ) 115 , 116 , EGFR 117–122 , HER2/Erb2 123 , HER4/ErbB4 124 , 125; and allosteric regulation of AKT 126–131 and PDK1 kinases 132 , 133 via docking of a phosphorylated hydrophobic motif to a hydrophobic pocket on the N-terminal lobe in the catalytic domain ., Activation processes in the ABL kinase is linked with the formation of multi-protein regulatory complexes with the SH2 and SH3 domains ., Crystallographic studies have determined that in the downregulated inactive state of the ABL-SH2-SH3 complex the SH3-SH2 unit is docked onto the kinase catalytic domain 107 , 108 ., In contrast , small angle X-ray scattering ( SAXS ) analysis has detected a dramatic structural rearrangement in the active ABL complex accompanied by the release of the inhibitory interactions and disengagement of the SH2-SH3 domains 107 , 108 ., Hydrogen exchange mass spectrometry ( HX MS ) investigation of the ABL-T315I dynamics has provided the first evidence of long-range conformational disturbances caused by activating mutations and allosterically transmitted to the remote protein regions 111 ., Recently discovered allosteric inhibitors GNF-2 , GNF-5 of the ABL kinase can bind to the myristoyl-binding pocket and independently inhibit kinase activity 112 , 113 ., HX MS studies of the ABL-T315I dynamics in the presence of ATP competitive inhibitor Dasatinib and GNF-5 have revealed long-range cooperativity between the myristate-binding site and the ATP-binding site induced upon allosteric inhibitor binding that allowed for the effective synergistic inhibition of the ABL-T315I mutant by a drug combination 114 ., Structure-functional studies of EGFR kinase domains have revealed that the formation of an asymmetric kinase dimer is critically associated with an activated kinase conformation and is essential for tyrosine kinase activation 117–122 ., A recent crystal structure of the HER2 kinase domain 123 has provided additional support to allosteric activation via asymmetric dimerization , similar to activation mechanisms in the EGFR and HER4 kinases 124 , 125 ., In common to the crystal structures of EGFR , HER2 and HER4 kinases , activation mechanisms may exploit an asymmetric head-to-tail dimer , in which the C- lobe of one monomer acts as a “donor” monomer ( activator ) that interacts with the N-lobe of an adjacent “acceptor” monomer ( receiver ) , stabilizing conformational changes that activate the receiver molecule ., Moreover , an asymmetric structural arrangement of a functional EGFR dimer is highly similar to the complex formed by the receiver CDK2 kinase with its activator , cyclin A 115 , 116 ., Recent studies have shown the importance of the intracellular juxtamembrane EGFR region in promoting activation of an asymmetric dimer via forming a “juxtamembrane latch” between the N-terminal lobe of the receiver and the C-terminal lobe of the activator , allowing to “glue” two kinase monomers and potentiate activation of the receiver molecule 120 , 121 ., Hence , a unifying structural mechanism associated with asymmetric tyrosine kinase arrangements in regulatory complexes could underlie the activation mechanism of the EGF and ErbB protein families and explain a linkage between ligand-induced receptor dimerization and kinase activation 134–137 ., Mechanisms of protein kinases regulation have been also studied in computational investigations of c-Src kinase 138–145 , adenylate kinase 146 , ABL kinase 147 , CDK5 kinase 148 , KIT kinase 149 , PKA kinase 150 , and AKT/PKB kinase 151 , RET and MET kinases 152 , 153 and EGFR kinase 154–160 ., These studies have suggested that functional coupling between collective motions and local structural changes can rationalize the experimental data and provide molecular insights into allosteric mechanisms ., We have previously reported that the impact of the gate-keeper mutant on conformational dynamics of ABL may spread far beyond the immediate site of mutation leading to functional changes in conformational mobility at the remote kinase regions 155 ., These results corroborated with the HX MS experiments of ABL regulatory complexes 111 , pointing to a potential allosteric effect of the activating mutations in the ABL kinase ., Despite recent progress in computational and experimental studies of protein kinase structure and function , the molecular mechanism and dynamics of mutation-induced allosteric kinase activation by regulatory complexes remain mostly qualitative ., In this work , we have investigated mechanistic aspects of allosteric activation mechanisms in ABL and EGFR kinases by integrating the results of multiscale simulations with the principal component analysis and computational modeling of signal propagation in proteins ., We show that mechanisms of allosteric activation in the ABL and EGFR kinases may be determined by a functional cross-talk between the organizing αF-helix and conformationally adaptive αI-helix and αC-helix ., These structural elements form a dynamic network of efficiently communicated clusters that may control the long-range coupling and allosteric activation in the interdomain regulatory complexes ., The results of study may reconcile current experimental data pointing to general mechanistic aspects of activating transitions in protein kinases ., In this work , molecular dynamics ( MD ) simulations , principal component analysis ( PCA ) and computational modeling of signal propagation in proteins 161 , 162 were employed to elucidate molecular principles of allosteric communication in ABL and EGFR kinases and determine allosteric signatures of the gate-keeper cancer mutations at the increasing level of complexity - from catalytic domain ( ABL-T315I , EGFR-T790M ) to multi-domain regulatory complexes ( ABL-T334I , EGFR-T766M ) ., The following specific objectives were pursued in the present study:, ( a ) perform a comparative analysis of the collective protein motions and allosteric communication profiles obtained from simulations of the catalytic domain and regulatory complexes;, ( b ) determine key structural elements and functional residues in ABL and EGFR kinases involved in collective motions and long-range allosteric coupling; ( d ) analyze and compare long-range communications and allosteric signatures of mutation-induced kinase activation by the gate-keeper mutations in ABL and EGFR ., We employed concepts of the absolute and relative Long Range Communication Capability ( LRCC ) associated with the protein residues in the context of a computational model of signal propagation in proteins 161 ( see Materials and Methods for a detailed description ) ., According to this model , two remote protein residues ( or residue clusters ) are defined to have a high communication propensity ( or communication capability ) if the mean-square fluctuation of their inter-residue distance would vary within a relatively small range over long time MD simulations ., The higher the fraction of residues that have high communication efficiency with a given residue at an empirically chosen threshold of efficient and fast long-range communications , the greater would be the absolute LRCC of this residue ., Under this assumption , a perturbation in one residue in a pair of well-communicated residues should be consistently “communicated” to the “partnering” residue located at a significant distance ., Conversely , two residues would not efficiently communicate when the thermal fluctuations of their inter-residue distance would be large and inconsistent with the level of displacements for respective residues , e . g . the inter-residue distance would change in a wide spectrum of values inconsistent with the amplitude of thermal fluctuations of individual residues ., This would amount to a slow and inconsistent propagation of a perturbation signal from one residue to the other ., A meaningful metric of long-range communication would display fluctuations of the inter-residue distance corresponding and reflecting the order of per residue fluctuations ., According to our model , mutational changes can modulate the energy landscape of a protein and alter communication propensities among pairs of residues ., The communication propensity histograms of the different protein states scan LRCC that represent the fraction of residues that may have high communication efficiency with a given residue and located at distances greater than a defined threshold from that residue ., We hypothesize that the residue clusters characterized by the peaks in the communication efficiency profile may be important for long-range cooperative interactions ., In this section , we analyzed collective motions and long-range communications in the ABL kinases using the results of our recently reported simulations of the ABL catalytic domain and multidomain complexes 155 in the following functional states: inactive ABL state ( PDB ID 1IEP ) 70 , the active ABL state ( PDB ID 1M52 ) 71 , 72; the active form of the ABL-T315I mutant ( PDB ID 2Z60 ) 75; the inactive autoinhibited form of ABL-SH2-SH3 complex ( PDB ID 2FO0 ) and the active form of the ABL complex ( PDB ID 1OPL ) 108 ., For clarity and completeness of the discussion , we summarized our earlier results of 20 ns MD simulations of the ABL complexes 155 in the context of the current objectives ( Figure S1 ) ., According to our results , a structurally stable bundle of α-helices in the C-terminal could be dynamically coupled via regulatory spines with the regions of larger thermal fluctuations corresponded to the P-loop , αC-helix and the activation loop ., Moreover , mutation-induced modulation of protein flexibility in the inactive state may be compounded by the increased structural stability of the active form 155 ., To better characterize the nature of collective motions between functional kinase regions in regulatory complexes , we analyzed here collective motions of the ABL complexes using PCA of the covariance matrix , calculated from 20 ns MD trajectories of the complete systems ( Figure S2 ) ., The correlation matrix described the linear correlation between any pairs of Cα atoms as they move around their average position during simulations ., A positive correlation between two atoms could reflect a concerted motion in the same direction , whereas a negative correlation may indicate an opposite direction motion ., We noted important similarities and differences between the correlation profiles of inactive ( Figure S2 A ) and active forms ( Figure S2 B ) ., The covariance map of the active ABL complex displayed an increased and more broadly distributed level of positive correlation , both within the catalytic core , and also between the catalytic domain and the SH2 domain ., Overall , the correlated motions along the first eigen mode in the active ABL complex represented a more uniform level of motion between sub-domains with lower amplitude fluctuations as compared to the autoinhibited form ., The “breathing” inter-lobe motion of the catalytic core was coupled with the motions of the activation loop , αC-helix and the αG-helix of the C-terminal ., The inter-lobe motions were also coupled with the collective inter-domain motions between the catalytic domain and the SH2 domain ., It is worth noting that the crystal structure of the active ABL complex ( PDB ID 1OPL ) is completely missing SH3 domain 108 ., Consequently , to ensure a consistent comparison of long-range communications between inactive and active complexes , we focused our analysis on the catalytic domain residues and a comparison of their communication profiles derived from simulations of the isolated catalytic core ( Figure, 1 ) and ABL complexes ( Figure 2 ) ., Structural mapping of long-range communications in the ABL catalytic domain was done using a reference distance of 30 Å ( Figures 1 A–C ) revealing a coupling between the N-terminal αC-helix ( residues 280–292 ) and a C-terminal core cluster comprised by the αF-helix ( residues 418–433 ) , αI-helix ( residues 486-496 ) , and αE-helix ( residues 337–356 ) ( Figure 1 ) ., According to our analysis , allosteric communication between a flexible β-strand of the N-terminal lobe , the αC-helix and the P+1 loop could be controlled by the integrating αF-helix ., The relative LRCC values between the inactive and active WT forms ( Figures 1 D–F ) reflected a partial loss in the communication capabilities in the active state and the increased mobility of the αF-helix and the αE-helix , indicative of a more flexible active kinase form ., We observed that this effect may be partly offset by a moderate increase in communication propensities of the hinge region ( residues 310–335 ) and the αC-helix ( Figure 1 D , E ) ., An interesting finding from this analysis was an appreciable impact of ABL-T315I on long-range communications , manifested in a broadened network of allosterically coupled residues ( Figures 1 D–F ) ., This effect could be seen by inspecting changes in the density of allosterically coupled clusters between the inactive states ( Figure 1A ) , the active form ( Figure 1B ) and the active form of the ABL-T315I mutant ( Figure 1C ) ., The αC-helix , αF-helix , and αI-helix regained their integrating role in enabling long-range communications in the mutant form ., A similar pattern of long-range communications between structurally rigid αF-helix and conformationally adaptive αI-helix , αC-helix and P+1 loop was detected from simulations of the ABL complexes ( Figures 2 A , B ) ., The impact of the ABL-T334I mutation on the inactive complex manifested in a decline of the LRCC values and the increased mobility of the αF-helix ( residues 437–453 ) , αE-helix ( residues 356–376 ) , αI-helix ( residues 511–528 ) , and αC-helix ( residues 299–311 ) ( Figures 2 A , C ) ., This observation is consistent with MD simulations of ABL complexes 155 , where ABL-T315I was shown to weaken the “rigid-clamp” arrangement and destabilize the inactive complex ., The efficient long-range communications between the organizing αF-helix and conformationally adaptive αI-helix and αC-helix is not merely a result of small thermal fluctuations in the respective segments ., A dynamic network of long-range interactions between these regions may rather have a functional significance in coordinating collective inter-lobe and inter-domain motions , both of which are known to be important to allosteric communication 35 ., A broader and denser network of long-range communicated clusters in the ABL-T315I included also the hinge region , the hydrophobic spine and the catalytically critical Asp-Phe-Gly ( DFG ) motif from the activation loop ( Figure 1C ) ., The critical role of the integrated cluster formed by the hydrophobic and catalytic spines , both anchored to the integrated αF-helix , is well recognized as an organizing element regulating protein kinase dynamics and activity 66–69 ., The cooperative interactions between the αF-helix and the αC-helix may control a dynamic connection between the two lobes of the catalytic core and be important for a dynamic assembly and disassembly of the hydrophobic spine regulating the protein kinase activity ., The combined analysis of the correlated motions and long-range communications in the ABL complexes is consistent with a mechanistic model of kinase activation involving cooperative assembly of the hydrophobic spine , the formation of the Src-like intermediate structure , and a cooperative breakage and formation of characteristic salt bridges 155 ., It is worth stressing that coupling between rigid and flexible protein regions and correlation of various motions may generally lead to both increases and decreases in thermodynamic stability ., A broader network of concerted motions and long-range communications in the mutant form is consistent with our previous finding that all free energy components may act concertedly to enhance the thermodynamic stability of the active ABL-T315I 155 ., Analysis of long-range communications allowed to highlight a functional role of stabilizing interdomain contacts in the inactive and active ABL complexes ( Figure 3 ) ., In the inactive ABL complex , the SH2 domain is docked closely onto the kinase domain by repositioning and rigidifying the αI-helix of the C-terminal and forming a dense network of hydrogen bonds and packing interactions ( Figure 3 A , B ) ., We observed high average occupancies for the major interdomain contacts that maintained their stability throughout a long simulation period ., These specific contacts included hydrogen bonding between side-chain of Arg-153 of the SH2 domain and the backbone carbonyls of the kinase domain residues Gln-517 and Glu-518 ., Additional hydrogen bonding was formed between Arg-189 of the SH2 domain and Asp-523 of the αI-helix ., This hydrogen bonding network was further strengthened by packing interactions between Tyr-158 of the SH2 domain , which aromatic ring was perfectly stacked against Tyr-361 from the αE-helix of the kinase domain ( Figure 3 B ) ., Importantly , the high occupancies of these interdomain contacts were significantly reduced for the ABL-T334I mutant ( Figure 3 C ) ., The interdomain interactions of the active ABL complex included Ile-164 of the SH2 domain interacting with Thr-291 and Tyr-331 of the kinase domain ., Interestingly , the occupancies of these core interactions were sustained in the active “top-hat” ABL complex at a relatively high level and even further consolidated for the mutant complex ( Figure 3 D ) ., In agreement with the experimental data , this analysis provided another evidence of a detrimental impact of the activating mutation on structural integrity of the inactive ABL complex that could promote conformational transformation to the active state ., Conversely , a stabilizing role of this mutation may be seen in the enhanced structural rigidity of the interface in the active ABL form ., Based on this analysis , we could suggest that the activating ABL-T334I mutation could perturb the interdomain interface via allosteric coupling between αF-helix and αI-helix and lead to a significant destabilization of the “rigid-clamp” form of the ABL-SH2-SH3 complex ., Hence , the impact of the gate-keeper mutation may be allosterically transmitted to the interdomain regions located at a considerable distance from the mutational site , supporting allosteric nature of the mutation-induced ABL activation ., Collectively , these factors could contribute to the mutation-induced allosteric effect that may perturb the thermodynamic equilibrium away from the inactive form towards alternative conformational states and thus serve as a catalyst of activation ., These results corroborated with the crystallographic and functional studies of the ABL-T315I mutant 74 , 75 confirming an activating nature of the gatekeeper mutation 76 ., These findings may have certain relevance in the context of drug resistance effects and design of ABL inhibitors ., Our results suggested that the ABL-T315I mutation could allosterically strengthen and coordinate distinctive structural elements of the kinase core , leading to the enhanced structural consolidation of the constitutively active kinase form ., As a result , design of ABL inhibitors binding to the active form of the enzyme would inevitably have to overcome competition from cellular ATP ., Novel ABL inhibitors of ABL-T315I that bind to the inactive conformation could experience weaker competition from ATP and may act by preventing kinase activation , rather than by inhibiting kinase activity directly ., The evidence of efficient long-range communications in active ABL complexes may be of importance given the rapidly growing interest in developing novel and specific kinase inhibitors inhibition targeting allosteric regions ., Indeed , our study may have specific implications in light of recent experimental studies of allosteric kinase inhibition and cooperativity between the myristate- and ATP-binding sites of ABL 113 , 114 ., HX MS analysis of ABL-T315I in the presence of Dasatinib and allosteric inhibitor GNF-5 demonstrated that binding in the myristate-binding site can elicit allosteric alterations in the conformational dynamics of the C-terminal αI-helix that are propagated to the β-strand of the C-terminal lobe and the ATP-binding site 114 ., The analysis of collective motions pointed to a possibility of concerted motions between the β-strand in the N-terminal lobe ( residues 260–280 in the inactive complex ) and the αI-helix from the C-terminal ( residues 446–463 in the inactive complex ) ( Figure S2 ) ., Interestingly , our analysis also suggested that allosteric coupling between a flexible β-strand of the N-terminal lobe , the αC-helix and the P+1 loop may be mediated and controlled by the integrating αF-helix ., Moreover , we found that the C-terminal αI-helix and the β-strand of the N-terminal lobe could be involved in the long-range communication of the down-regulated ABL complex and allosteric coupling of these functionally important binding sites could be modulated by the gate-keeper mutation ., The importance of these results may be appreciated in the context of experimentally detected allosteric effect of the GNF-5 inhibitor that binds to the myristate-binding site and can allosterically affect the thermodynamic stability of the ATP-binding site residues from the β-strand 113 , 114 ., Hence , our results are in accordance with a mechanistic view of allosteric ABL activation emerging from the experimental data ., We propose that allosteric inhibitor binding with ABL-T315I may lead to concerted changes of conformational mobility in these regions , thereby restoring structural arrangement of the ATP-binding site compatible with Dasatinib binding ., A detailed analysis of allosteric ABL inhibition by small molecules is being currently pursued in conjunction with the experimental verification by our collaborators , which a subject of a separate investigation that extends beyond the scope of the current study and will be presented elsewhere ., In this section , we analyzed allosteric signatures of the EGFR kinase catalytic domain using the results of MD simulations in the following functional states: the inactive EGFR form ( PDB ID 1XKK ) 77; the active EGFR form ( PDB ID 2J6M ) 78 , the active form of the EGFR-T790M mutant ( PDB ID 2JIT ) 79 ., The analysis of long-range communications the EGFR catalytic domain revealed similar coupling between structurally rigid αF-helix and conformationally adaptive αI-helix , αC-helix of the catalytic core ( Figure 4 ) ., This effect was seen from inspecting changes in the distribution of communicated residue clusters in the inactive state ( Figure 4A ) , the active form ( Figure 4B ) and the active form of the EGFR-T790M mutant ( Figure 4C ) ., Mutation-induced amplification of protein flexibility in the inactive state could be accompanied by the counter-effect of restoring structural stability of the active mutant form ( Figures 4 D-F ) ., We found that structural elements of the catalytic core involved in long-range communications may be common in ABL and EGFR , e . g . structural architecture of the | Introduction, Results/Discussion, Materials and Methods | The emerging structural information about allosteric kinase complexes and the growing number of allosteric inhibitors call for a systematic strategy to delineate and classify mechanisms of allosteric regulation and long-range communication that control kinase activity ., In this work , we have investigated mechanistic aspects of long-range communications in ABL and EGFR kinases based on the results of multiscale simulations of regulatory complexes and computational modeling of signal propagation in proteins ., These approaches have been systematically employed to elucidate organizing molecular principles of allosteric signaling in the ABL and EGFR multi-domain regulatory complexes and analyze allosteric signatures of the gate-keeper cancer mutations ., We have presented evidence that mechanisms of allosteric activation may have universally evolved in the ABL and EGFR regulatory complexes as a product of a functional cross-talk between the organizing αF-helix and conformationally adaptive αI-helix and αC-helix ., These structural elements form a dynamic network of efficiently communicated clusters that may control the long-range interdomain coupling and allosteric activation ., The results of this study have unveiled a unifying effect of the gate-keeper cancer mutations as catalysts of kinase activation , leading to the enhanced long-range communication among allosterically coupled segments and stabilization of the active kinase form ., The results of this study can reconcile recent experimental studies of allosteric inhibition and long-range cooperativity between binding sites in protein kinases ., The presented study offers a novel molecular insight into mechanistic aspects of allosteric kinase signaling and provides a quantitative picture of activation mechanisms in protein kinases at the atomic level . | Despite recent progress in computational and experimental studies of dynamic regulation in protein kinases , a mechanistic understanding of long-range communication and mechanisms of mutation-induced signaling controlling kinase activity remains largely qualitative ., In this study , we have performed a systematic modeling and analysis of allosteric activation in ABL and EGFR kinases at the increasing level of complexity - from catalytic domain to multi-domain regulatory complexes ., The results of this study have revealed organizing structural and mechanistic principles of allosteric signaling in protein kinases ., Although activation mechanisms in ABL and EGFR kinases have evolved through acquisition of structurally different regulatory complexes , we have found that long-range interdomain communication between common functional segments ( αF-helix and αC-helix ) may be important for allosteric activation ., The results of study have revealed molecular signatures of activating cancer mutations and have shed the light on general mechanistic aspects of mutation-induced signaling in protein kinases ., An advanced understanding and further characterization of molecular signatures of kinase mutations may aid in a better rationalization of mutational effects on clinical outcomes and facilitate molecular-based therapeutic strategies to combat kinase mutation-dependent tumorigenesis . | algorithms, physics, computer science, computational chemistry, computer modeling, statistical mechanics, chemistry, biology, computational biology, biophysics | null |
journal.pntd.0003261 | 2,014 | Whole Genome Sequence of the Treponema pallidum subsp. endemicum Strain Bosnia A: The Genome Is Related to Yaws Treponemes but Contains Few Loci Similar to Syphilis Treponemes | Uncultivable human pathogenic treponemes include T . pallidum subsp ., pallidum ( TPA ) , causing syphilis , T . pallidum subsp ., pertenue ( TPE ) , causing yaws , and T . pallidum subsp ., endemicum ( TEN ) , causing bejel , which is also known as endemic or nonvenereal syphilis ., Infections caused by TPE and TEN are commonly denoted as endemic treponematoses ., While yaws is found in warm , moist climates , bejel is found in drier climates ., In both cases , infection is spread by direct contact ( e . g . skin-to-skin or skin-to-mucosa ) ., In addition , bejel can also be transmitted by contact with contaminated utensils 1 , 2 ., The current , and widespread , belief that yaws and bejel are non-sexually transmitted may simply reflect that these diseases mostly affect children that have not reached sexual maturity 3 , 4 ., Diagnosis of endemic treponematoses comprises clinical symptoms , epidemiological data , and serology ., Since there is significant clinical similarity between the symptoms of syphilis and endemic treponematoses , and serology cannot discriminate between infection with TPA , TPE , and TEN strains , the epidemiology plays a major role in establishing a diagnosis ., While yaws remains endemic in poor communities in Africa , Southeast Asia , and the western Pacific , bejel is predominant in western Africa and in the Middle East ( reviewed in 2 , 4 ) ., Imported cases of yaws and bejel have been documented in children in Europe and Canada 5 , 6 ., With the accumulation of genetic data , molecular targets that can be used to differentiate treponemal subspecies , at the molecular level , have become available 2 ., Endemic syphilis has been described almost everywhere in Europe since the 16th century ( for review see 7 ) and often was described under different names , e . g . the disease that appeared in Brno , CZ in 1575 was called morbus Brunogallicus , although it is not clear whether this infection was not perhaps caused by the syphilis treponeme 8 ., The Bosnia A strain was isolated in 1950 in Bosnia , a country in southern Europe , from a 35-year old male with mucous patches under the tongue and on the tonsils; additionally , the patient showed secondary lesions ( papules ) on the face , trunk and extremities ., Material for experimental inoculation of laboratory animals was taken from an ulcer on the shaft of the penis 9 ., Although several other isolates were collected from bejel patients , only one additional strain of T . pallidum subsp ., endemicum ( Iraq B ) is currently propagated in laboratory settings ., In this study , the complete genome sequence of the T . pallidum subsp ., endemicum Bosnia A strain was obtained using a combination of next-generation sequencing approaches and compared to the genomes of the four TPE strains ( Samoa D , CDC-2 , Gauthier , Fribourg-Blanc isolate ) and five TPA strains ( Nichols , DAL-1 , Chicago , SS14 , Mexico A ) , all of which have been determined in recent years 10–15 ., Bosnia A DNA was provided by Dr . Sylvia M . Bruisten from the Public Health Service , GGD Amsterdam , Amsterdam , The Netherlands ., Bosnia A genomic DNA was amplified using the pooled segment genome sequencing ( PSGS ) method as described previously 11 , 15 ., Briefly , Bosnia A DNA was amplified with 214 pairs of specific primers to obtain overlapping PCR products ( Table S1 ) ., To facilitate sequencing of paralogous genes containing repetitive sequences , PCR products were mixed in equimolar amounts into four distinct pools ., Prior to next-generation sequencing ( 454-pyrosequencing , Illumina and SOLiD ) , the PCR products constituting each pool were labeled with multiplex identifier ( MID ) adapters and sequenced as four different samples ., Two genomic regions were not amplified during PSGS and therefore were not used for sequencing the whole genome ( gaps between coordinates 332290–335395 and 1123251–1123648 according to the Nichols sequence , AE000520 . 1 16; see Table S1 ) ., Sequences in these regions were Sanger sequenced at the University of Washington in Seattle ( WA ) , USA ., Whole genome DNA sequencing was done using the Applied Biosystems/SOLiD 3 System platform ( Life Technologies Corporation , Carlsbad , CA , USA ) combined with the Roche/Genome Sequencer FLX Titanium platform ( 454 Life Sciences , Branford , CT , USA ) and with the Illumina/Solexa HiSeq 2000 approach ( Illumina , San Diego , CA , USA ) ., SOLiD sequencing was performed at SeqOmics Ltd ( Mórahalom , Hungary ) , 454-pyrosequencing and Illumina sequencing were performed at The Genome Institute , Washington University School of Medicine ( St . Louis , MO , USA ) ., SOLiD , 454 , and Illumina sequencing resulted in average read lengths of 40 bp , 504 bp and 100 bp and the total average depth coverage of 234× , 138× and 141× , respectively ., 454 and Illumina sequencing reads were obtained from 4 distinct pools ( sequenced as 4 different samples – see Table S1 ) and were separately assembled de novo using a Newbler assembler ( 454 Life Sciences , Branford , CT , USA ) or TIGRA 17 , respectively ., The resulting 454 and Illumina contigs obtained for each pool were then aligned to the corresponding sequences ( representing each pool sequence ) of the reference CDC-2 genome ( CP002375 . 1 11 ) using Lasergene software ( DNASTAR , Madison , WI , USA ) ., All gaps and discrepancies between these platforms within each pool were resolved using Sanger sequencing ., Altogether , 20 genomic regions of the Bosnia A genome were amplified and Sanger sequenced ., The final overlapping pool sequences were joined to obtain complete genome sequence of the Bosnia A strain ., The SOLiD sequencing results were mapped to the reference Samoa D genome ( CP002374 . 1 11 ) using the CLC Genomics Workbench ( CLC bio , Cambridge , MA , USA ) and were processed as mentioned above ., The genome sequence obtained from SOLiD was then compared with the consensus genome sequence obtained from 454 and Illumina ., All discrepancies were resolved using Sanger sequencing ., Two TPE genomes ( CDC-2 or Samoa D ) were used as reference genomes for contig alignments since only few minor genetic differences have been found to be specific within individual TPE strains 11 ., Due to low coverage , one genomic region ( Treponema pallidum interval; TPI ) , was amplified with specific primers using a GeneAmp XL PCR Kit ( Applied Biosystems , Foster City , CA , USA ) 18 , 19 ., This TPI-48 interval contained paralogous genes tprI and tprJ ., The PCR product was purified using a QIAquick PCR Purification Kit ( QIAGEN , Valencia , CA , USA ) according to the manufacturers instructions and Sanger sequenced using internal primers ., The tprK ( TENDBA_0897 ) , arp ( TENDBA_0433 ) , and TENDBA_0470 genes were amplified and cloned into the pCR 2 . 1-TOPO cloning vector ( Invitrogen , Carlsbad , CA , USA ) ., Nine independent clones for the tprK and arp genes and seven clones for TENDBA_0470 were sequenced as previously described 11 ., A total of 7 genomic regions ( in genes TENDBA_0040 , TENDBA_0348 , TENDBA_0461 , TENDBA_0697 , TENDBA_0859 , TENDBA_0865 and TENDBA_0966 ) revealed intra-strain variability in the length of homopolymeric ( G- or C- ) stretches ., The prevailing length of these regions was determined by TOPO TA-cloning and Sanger sequencing ., At least five independent clones were sequenced as previously described 15 ., The final whole genome sequence of the Bosnia A strain was assembled from SOLiD , 454 and Illumina contigs ., In addition , Sanger sequencing was used for finishing the complete genome sequence and for additional sequencing including paralogous , repetitive and intra-strain variable chromosomal regions ., Geneious software v5 . 6 . 5 20 was used for gene annotation based on the annotation of the TPE CDC-2 genome 11 ., Genes were tagged with TENDBA_ prefix ., The original locus tag numbering corresponds to the tag numbering of orthologous genes annotated in the TPE CDC-2 genome 11 ., The TENDBA_0897 gene , coding for TprK , showed intra-strain variable nucleotides and therefore nucleotides in variable regions were denoted with Ns in the complete Bosnia A genome ., For proteins with unpredicted functions , a gene size limit of 150 bp was applied ., Protein domains and functional annotation of analyzed genes were characterized using Pfam 21 , CDD 22 and KEGG 23 databases ., Whole genome nucleotide alignments of five TPA strains , four TPE strains and the Bosnia A strain were used for determination of genetic relatedness using several approaches including calculation of nucleotide diversity ( π ) and construction of a phylogenetic tree ., All positions containing indels in at least one genome sequence were omitted from the analysis ., There were a total of 1 , 128 , 391 nucleotide positions aligned in the final dataset ., TPA strains comprised Nichols ( re-sequenced genome CP004010 . 2 14 ) , DAL-1 ( CP03115 . 1 13 ) , SS14 ( re-sequenced genome CP004011 . 1 14 ) , Chicago ( CP001752 . 1 10 ) , and Mexico A ( CP003064 . 1 12 ) genomes , while TPE strains included Samoa D ( CP002374 . 1 11 ) , CDC-2 ( CP002375 . 1 11 ) , Gauthier ( CP002376 . 1 11 ) and Fribourg-Blanc ( CP003902 . 1 15 ) ., Whole genome alignments were constructed using Geneious software 20 and SeqMan software ( DNASTAR , Madison , WI , USA ) ., Nucleotide differences among studied whole genome alignments were analyzed using DnaSP software , version 5 . 10 24 ., An unrooted phylogenetic tree was constructed from the whole genome sequence alignment using the Maximum Parsimony method and MEGA5 software 25 ., To test , whether the mosaic character of identified loci were a result of intra-strain recombination , potential donor sites were screened from the entire Bosnia A genome using several computer programs and algorithms including RDP3 26 , EditSeq software ( DNASTAR , Madison , WI , USA ) , BLAST ( http://blast . ncbi . nlm . nih . gov ) , and Crossmatch ( http://www . phrap . org/phredphrapconsed . html ) ., We failed to find any potential donor sites in the Bosnia A genome ., We also failed to find any TPA- or TPE-specific NGS reads in the regions having a mosaic character ., The complete genome sequence of the Bosnia A strain was deposited in the GenBank under accession number CP007548 ., Sequencing of the TEN Bosnia A strain genome using three independent next-generation sequencing platforms yielded a total combined average coverage of 513× ., The summarized genomic features of the Bosnia A strain in comparison to previously sequenced TPA and TPE strain genomes are shown in Table, 1 . The size of the Bosnia A genome ( 1 , 137 , 653 bp ) was 1 , 628–2 , 828 bp shorter than the sizes of previously published genomes for TPA and TPE strains 10–15 ., The overall gene order in the Bosnia A genome was identical to other TPE and TPA strains ., Altogether , 1125 genes were annotated in the Bosnia A genome including 54 untranslated genes encoding rRNAs , tRNAs and other ncRNAs ( short bacterial RNA molecules that are not translated into proteins ) ., A total of 640 genes ( 56 . 9% ) encoded proteins with predicted function , 137 genes encoded treponemal conserved hypothetical proteins ( TCHP , 12 . 2% ) , 141 genes encoded conserved hypothetical proteins ( CHP , 12 . 5% ) , 145 genes encoded hypothetical proteins ( HP , 12 . 9% ) and 8 genes ( TENDBA_0082a , TENDBA_0146 , TENDBA_0316 , TENDBA_0370 , TENDBA_0520 , TENDBA_0532 , TENDBA_0812 and TENDBA_1029; 0 . 7% ) were annotated as pseudogenes ., The average and median gene lengths of the Bosnia A genome were calculated to 979 . 2 bp and 831 bp , respectively ., The intergenic regions covered 52 . 6 kbp and represented 4 . 63% of the total Bosnia A genome length ., In general , other calculated genomic parameters were similar to other TPE strains ., When compared to TPA strains , the Bosnia A genome contained a 635 bp long insertion in the tprF locus ., In this respect , the Bosnia A genome was similar to TPE strains ., When compared to both TPA and TPE genomes , the Bosnia A genome contained a 2300 bp long deletion involving the tprF and G loci ( TPANIC_0316 and TPANIC_0317 in the Nichols genome CP004010 . 2 14 ) ., Moreover , the predicted TENDBA_0316 gene ( 1860 bp in length ) was a chimera encompassing the tprG 5′-region , tprI-like sequence and the tprF 3′-region , and was hence designated as tprGI as previously described by Centurion-Lara et al . 27 ( Table 2 ) ., Two insertions of 65 bp and 52 bp , respectively , resulted in the prediction of two hypothetical genes , TENDBA_ 0126b and TENDBA_548a ., The same orthologs were also predicted in TPE but not in TPA strains ( Table 2 ) ., Besides the annotated pseudogenes in the Bosnia A genome ( see above ) , 8 additional genes ( orthologous to TP0129 , TP0132 , TP0135 , TP0266 , TP0318 , TP0370 , TP0671 and TP1030 ) were considered pseudogenes ., The same genes were also considered pseudogenes in TPE strains 11 , 15 ( Table 1 ) ., Sequence relatedness of the Bosnia A genome to other Treponema pallidum genomes is shown in Fig ., 1 . This unrooted tree was constructed using several available whole genome sequences of uncultivable pathogenic treponemes ., The image clearly showed clustering of the Bosnia A strain with the TPE strains ., The Bosnia A genome was found to be 99 . 91–99 . 94% and 99 . 79–99 . 82% identical to the TPE and TPA genomes , respectively ( Table 3 ) ., The nucleotide diversity between TPE strains and the Bosnia A strain ( 0 . 00063±0 . 00032 to 0 . 00086±0 . 00043 ) was about three times lower than the nucleotide diversity between TPA strains and the Bosnia A strain ( 0 . 00181±0 . 00090 to 0 . 00212±0 . 00106 ) ., For comparison , calculated π values between the Bosnia A strain and individual TPA strains were of the same order of magnitude as π values between TPA and TPE strains ( Table 4 ) ., To identify Bosnia A-specific differences , the Bosnia A genome was compared to the available genomes of TPE strains 11 , 15 and TPA strains 10 , 12–14 ., The Bosnia A strain-specific sequences were defined as those not present in both TPA and TPE strains and altogether comprised 406 differences ( indels and substitutions with a total length of 2772 bp ) equally distributed along the Bosnia A genome ( Fig . 2 ) ., Differences in coding regions included 9 deletions , 5 insertions and 360 nucleotide substitutions for a total of 2728 bp ( Table 5 ) ., Those 360 substitutions resulted in 197 Bosnia A-specific amino acid differences in the putative proteome ., Most of the nucleotide substitutions were found in the TENDBA_0136 , TENDBA_0548 , TENDBA_0856 , TENDBA_0859 and TENDBA_0865 genes ( Table 5 ) ., Bosnia A-specific frameshift mutations ( caused by three deletions and one insertion ) resulted in significant gene truncation ( TENDBA_0082a , TENDBA_0316 and TENDBA_1029 ) or elongation ( TENDBA_0126b ) ( Table 2 ) ., Other detected indels resulted in 6 protein shortenings ( TENDBA_0067 , TENDBA_0136 , TENDBA_0225 , TENDBA_0548 , TENDBA_0859 , and TENDBA_0865 ) and 4 protein elongations ( TENDBA_0856 , TENDBA_0859 , TENDBA_0897 , and TENDBA_0898 ) ( Table 5 ) ., All affected genes code for hypothetical proteins of unknown function except for TENDBA_0898 coding for RecB ( exodeoxyribonuclease V beta subunit; EC3 . 1 . 11 . 5 ) ., TENDBA_0136 and TENDBA_0865 have been predicted to be putative outer membrane proteins ., In addition , TPA and TPE orthologs to TENDBA_0136 have been experimentally shown to bind human fibronectin 28 ., TENDBA_0856 has been predicted to be putative lipoprotein ., No putative conserved domains have been detected in hypothetical proteins except for TENDBA_0067 , TENDBA_0225 and TENDBA_1029 containing TPR ( tetratricopeptide ) domain , LRR_5 ( leucine rich repeat ) domain and DbpA ( RNA binding ) domain , respectively ( Table 5 ) ., All nonsynonymous substitutions have been identified outside the predicted domains ., Genome sequences differentiating the Bosnia A strain from the TPA but not TPE strains are shown in Fig ., 2 . These sequences were found to be regularly distributed along the Bosnia A genome and altogether comprised 1422 differences ( indels and substitutions of total length of 2335 bp ) ., In the coding regions , 2128 bp including 13 deletions , 9 insertions and 1296 substitutions differentiated genomes of TPA strains from Bosnia A and other TPE strains ( Table 6 ) ., A set of 1296 substitutions resulted in 631 amino acid differences in the encoded proteins ., Most of the differences were found in genes TENDBA_0117 ( tprC ) , TENDBA_0131 ( tprD ) , TENDBA_0133 , TENDBA_0134 , TENDBA_0136 , TENDBA_0304 , TENDBA_0314 , TENDBA_0462 , TENDBA_0619 , TENDBA_0620 ( tprI ) , and TENDBA_0621 ( tprJ ) ( Table 6 ) ., Except for TENDBA_0103 coding for RecQ ( ATP-dependent DNA helicase; EC3 . 6 . 4 . 12 ) and TENDBA_0027 coding for HlyC ( putative hemolysin ) , all other affected genes code for hypothetical proteins of unknown function ., TENDBA_0134 has been predicted to be putative outer membrane protein ., TENDBA_0462 and TENDBA_0858 have been predicted to be putative lipoproteins ., No putative conserved domains have been detected in hypothetical proteins except for TENDBA_0067 and TENDBA_0304 conatining TPR ( tetratricopeptide ) domain and peptidase_MA_2 domain , respectively ( Table 6 ) ., All nonsynonymous substitutions have been identified outside the predicted domains ., Genome sequences differentiating the Bosnia A strain from TPE but not TPA strains are shown in Fig ., 2 . These sequences were also found to be regularly distributed along the Bosnia A genome and , altogether , comprised 197 differences in genome positions ( containing indels and substitutions encompassing a total of 635 bp ) ., Three deletions , three insertions and 174 substitutions ( Table 7 ) were found within the Bosnia A coding regions , encompassing a total of 612 bp ., The 174 substitutions resulted in 101 amino acid differences in the putative encoded proteins ., Most of the substitution differences were found in genes TENDBA_0136 , TENDBA_0488 , TENDBA_0577 , TENDBA_0856a/TENDBA_0858 , TENDBA_0859 , TENDBA_0865 and TENDBA_0968 ( Table 7 ) ., An insertion of 378 bp in TENDBA_1031 ( tprL ) resulted in a gene elongation ( Table 2 ) ., TENDBA_0488 codes for Mcp ( methyl-accepting chemotaxis ) protein ., All other genes code for hypothetical proteins of unknown function ., Two genes have been predicted to encode putative outer membrane proteins ( TENDBA_0136 and TENDBA_0865 ) and one gene has been predicted to encode putative lipoprotein ( TENDBA_0858 ) ., No putative conserved domains have been detected in hypothetical proteins ( Table 7 ) ., Despite the overall sequence similarity of the Bosnia A genome to TPE strains , several chromosomal sequences were found to be almost identical to sequences in TPA strains ., The Bosnia A sequence in the TENDBA_0577 locus was identical to four out of 5 orthologous sequences of completely sequenced TPA strains ( Fig . 3 ) ., In the TENDBA_0968 locus , stretches of TPA- and TPE-like sequences were found ( Fig . 3 ) and a similar pattern was also found in TENDBA_0858 ( not shown ) ., In addition , TENDBA_0326 ( tp92 , bamA ) was identical to the orthologous sequence of TPA SS14 ( coordinates 1593–1649 , Fig . 3 ) and to all TPA strains ( with the exception of the TPA Mexico A strain ) between coordinates 2127–2494 ., The TPA Mexico A strain is , in this region , similar to TPE strains 12 , 29 ., While the latter TPA-like sequences in TENDBA_0326 were almost 0 . 4 kbp long , other TPA-like sequences were usually relatively short , ranging from about 50–70 bp ., However , TPA-like sequences of the Bosnia A strain were clearly different from Bosnia A-specific sequences with sporadic nucleotide positions identical to TPA sequences ( TENDBA_0856; Fig . 3 ) ., The previously reported 378 bp insertion almost identical to TPA strains ( differing only in one nucleotide position 27 ) was confirmed in TENDBA_1031 as well as the nucleotide mosaic in the TP0488 ( mcp2-1 ) locus; revealing a sequence identical to TPA Mexico A ( with the exception of 2 single nucleotide substitutions 12 ) ., Altogether , at least seven TPA-like sequences having 5 or more nucleotide positions identical to TPA sequences and not interrupted by TPE-like nucleotide positions were found in the Bosnia A genome ., The first complete genome sequence of the bejel-causing agent , T . pallidum subsp ., endemicum ( TEN ) strain Bosnia A , was determined using three independent next-generation sequencing techniques ., Because the total combined coverage was >500× and all sequencing ambiguities were resolved with Sanger sequencing , the quality of this new genome is very high ., This allowed us to carry out a comparative analysis of the Bosnia A genome with the already available treponemal genomes 10–15 , 30 with a high degree of confidence that our results would not be affected by sequencing errors ., In several of the previously published genomes , the whole genome sequence was compared to whole genome fingerprinting data to assess the quality of the genome sequence ., In each of the previously tested genomes , the sequencing error rate was less than 10−4 11 , 12 , 15 , 30 ., The genome length of strain Bosnia A ( 1 , 137 , 653 bp ) is about 2 kbp shorter than the length of TPE or TPA genomes ., This is caused by a 2300 bp deletion in the tprF and tprG loci ., This deletion was also confirmed in the TEN Iraq B sequence 27 suggesting that this is a common feature of bejel strains ., An identical deletion was also found in the T . paraluisleporidarum ec ., Cuniculus genome ( formerly denoted T . paraluiscuniculi Cuniculi A 30 , 31 ) ., Moreover , this type of deletion was observed during PCR amplification of the tprF and tprG loci in other treponemal genomes ( M . Strouhal , D . Šmajs; unpublished data ) ., This fact , together with the presence of repeats in the flanking regions suggests that this 2300 bp deletion is a result of polymerase slippage and that this deletion could have happened several times independently during evolution ., In fact , no other similarities between the Bosnia A and T . p ., ec ., Cuniculus genome were found with respect to other identified indels in the T . p ., ec ., Cuniculus genome ., The overall genetic similarity of Bosnia A to the sequenced TPE strains is 99 . 91–99 . 94% , at the DNA level ., For comparison , the sequence similarity between TPA and TPE strains is greater than 99 . 8% 11 , 15 ., This enormous sequence similarity among TPA , TPE and TEN strains is the molecular basis for the long established fact that individual etiological agents of syphilis and endemic treponematoses ( yaws and bejel ) cannot be distinguished by their morphology or serology ., Although syphilis , yaws , and bejel show differences in their geographical distribution , mode of transmission , invasiveness and pathogenicity , it is known that the clinical symptoms of these diseases overlap and one disease can mimic the others ., Interestingly , in very dry areas , yaws symptoms are almost the same as bejel symptoms 32; which again reflects the extremely high sequence similarity between TPE and TEN strains ., In many or perhaps most cases , the final diagnosis is therefore often based on the epidemiological context of the infection ., However , at the same time , even small genomic differences ( although not known at present ) have the potential to influence the phenotypic differences between the clinical manifestations of syphilis , yaws and bejel ., Additional whole genome sequences of TPA , TPE and TEN strains will help to identify a set of invariant differences between the etiological agents of these diseases , which could help answer this question ., At the same time , the TEN Bosnia A strain is clearly distant from the cluster of TPE strains ., However , additional TEN whole genome sequences will be needed to assess the variability within TEN strains ., To our knowledge , there is only one additional laboratory stock of TEN , i . e . strain Iraq B . Previous studies on the Iraq B isolate revealed a high degree of similarity to Bosnia A 27 , 29 , 33–36 suggesting that this strain is more related to Bosnia A than to TPE strains ., Most prominent genetic changes between Bosnia A and TPE and/or TPA genomes resulting in protein truncations or elongations were located in just 14 genes ., These genes encoded TprA , F , G , and L proteins , RecQ protein , ethanolamine phosphotransferase , and treponemal conserved hypothetical proteins ( 3 ) or hypothetical proteins ( 5 ) ., Both Tpr and RecQ proteins were found to also be affected in the T . p ., ec ., Cuniculus genome 30 ., While the tprA gene was functional in Bosnia A and TPE strains but not among TPA strains ( except for strain Sea 81-4; see 37 ) , tprF and tprG were partially deleted ( similarly to T . p . ec . Cuniculus genome ) and the tprL gene was elongated in a way that was similar to that seen in TPA strains ., These changes were already described in detail by Centurion-Lara et al . 27 ., Tpr proteins likely play an important role in treponemal infectivity , pathogenicity , immune evasion and host specificity ., Tpr proteins induce an antibody response during infection and exhibit heterogeneity both within and among T . pallidum subspecies and strains 38–40 ., In the T . p ., ec ., Cuniculus genome , a mutation in recQ resulted in a predicted RecQ protein without a C-terminal or DNA-binding domain 41; on the other hand in Bosnia A the frameshift reversion led to a functional recQ gene ( similar to that seen in TPE genomes 11 ) ., Other prominent changes seen in the Bosnia A strain include a different number of tandem repeat units in TENDBA_0433 ( encoding Arp ) and TENDBA_0470 genes ( encoding conserved hypothetical protein ) compared to orthologous genes in individual TPE and TPA strains ., The same number of 60-bp tandem repeat units ( all of Type II ) within the arp gene was found in the Bosnia A genome as previously described 42 ., Variable numbers of tandem repeat units in genes orthologous to TENDBA_0470 have already been described in TPE and TPA strains 11 , 15 , 19 ., The genome of Bosnia A showed several genetic loci with sequences identical to TPA sequences ( Fig . 3 ) ., The TENDBA_0577 gene encoded treponemal conserved hypothetical protein of unknown function with predicted cytoplasmic membrane localization ., This gene was completely identical to TPA orthologs and differed from TPE orthologs by deletion of 12 nucleotides and substitution of 5 nucleotides ., Recent studies of σ factor RpoE ( TP0092 ) binding sites identified gene TP0577 ( orthologous to TENDBA_0577 ) as one of 22 putative TP0092-controlled ORFs 43 ., The TENDBA_0577 thus could possibly code for a protein integrated in the stress response pathway during the first days post infection ., Similarly , the 378 bp insertion in TENDBA_1031 is with exception of a 1 nucleotide insertion almost identical to orthologs of the TPA strain ( but not to TPE strains ) ., In other genes ( TENDBA_0968 , TENDBA_0858 ) , 50–70 bp long sequences identical to one or several TPA strains were found indicating that the genome of Bosnia A incorporated sequences identical to TPA strains ., Most of the above mentioned genes were found to evolve under positive selection in TPA-TPE comparisons 11 ., In fact , previous papers found this type of mixed TPA and TPE sequences in TPA Mexico A and South Africa strains 12 , 29 ., Moreover , previous reports have shown that TEN strain Bosnia A contains the same nucleotide mosaic at the TP0488 ( mcp2-1 ) locus as TPA Mexico A ( with the exception of 2 single nucleotide substitutions ) ., Despite the numerous efforts to identify potential donor sites within TPA Mexico A that could explain the existence of these sequences by intra-strain recombination 12 , no such sites have been identified in the Mexico A genome ., Similarly , no donor sites have been identified in the Bosnia A genome either ., It is likely that these sequences identical to TPA in the Bosnia A genome could result from inter-strain recombination event between TPA and TEN strains during a simultaneous infection of multiple hosts during the TEN evolution ., Although the overall genome sequence of Bosnia A is related to TPE strains , horizontal gene transfer appears to be the mechanism that introduced at least seven chromosomal sequences related to TPA SS14 , TPA Mexico A , and other TPA strains ., In fact , both the TPA SS14 and Mexico A sequences are required and sufficient to provide sequences to Bosnia A genome ., Moreover , at least two subsequent transfers had to occur to introduce both SS14- and Mexico A-specific sequences ., Experimental infection with either TPA , TPE or TEN strains did not result in complete cross-protection 9 ., In addition , recombination mechanisms are more active during treponemal infection and represent important genetic mechanisms for avoiding the host immune response 40 ., Moreover , the absence of modification and restriction systems and the presence of genes for homologous recombination in pathogenic treponemes 16 appear to allow incorporation of foreign DNA molecules with subsequent integration into chromosomal DNA ., Therefore , uptake of TPA DNA by a TEN strain during a simultaneous infection of multiple hosts appears to be a possible explanation ., It is clear that TPA strains can be classified as SS14-like ( SS14 , Mexico A ) and Nichols-like strains ( Nichols , DAL-1 , Chicago ) 14 , 44 and that most of the TPA strains causing infections throughout the world are in fact SS14-like strains 36 ., However , it is not clear if the SS14 and Mexico A sequences in the Bosnia A genome reflect a greater prevalence of SS14-like strains in the human population or an accidental coincidence of transfers from SS14-like strains ., Moreover , there are several loci in the Bosnia A genome similar to the TENDBA_0856 locus ( TENDBA_0483 , TENDBA_0858 , TENDBA_0865 ) that represent regions of Bosnia A-specific sequences with only sporadic nucleotide positions that are identical to TPA sequences ., These sequences may be identical to other , yet unidentified , TPA strains or isolates ., If such TPA isolates are identified in the future , they may help to unravel the evolution of TPA and TEN treponemes . | Introduction, Materials and Methods, Results, Discussion | T . pallidum subsp ., endemicum ( TEN ) is the causative agent of bejel ( also known as endemic syphilis ) ., Clinical symptoms of syphilis and bejel are overlapping and the epidemiological context is important for correct diagnosis of both diseases ., In contrast to syphilis , caused by T . pallidum subsp ., pallidum ( TPA ) , TEN infections are usually spread by direct contact or contaminated utensils rather than by sexual contact ., Bejel is most often seen in western Africa and in the Middle East ., The strain Bosnia A was isolated in 1950 in Bosnia , southern Europe ., The complete genome of the Bosnia A strain was amplified and sequenced using the pooled segment genome sequencing ( PSGS ) method and a combination of three next-generation sequencing techniques ( SOLiD , Roche 454 , and Illumina ) ., Using this approach , a total combined average genome coverage of 513× was achieved ., The size of the Bosnia A genome was found to be 1 , 137 , 653 bp , i . e . 1 . 6–2 . 8 kbp shorter than any previously published genomes of uncultivable pathogenic treponemes ., Conserved gene synteny was found in the Bosnia A genome compared to other sequenced syphilis and yaws treponemes ., The TEN Bosnia A genome was distinct but very similar to the genome of yaws-causing T . pallidum subsp ., pertenue ( TPE ) strains ., Interestingly , the TEN Bosnia A genome was found to contain several sequences , which so far , have been uniquely identified only in syphilis treponemes ., The genome of TEN Bosnia A contains several sequences thought to be unique to TPA strains; these sequences very likely represent remnants of recombination events during the evolution of TEN treponemes ., This finding emphasizes a possible role of repeated horizontal gene transfer between treponemal subspecies in shaping the Bosnia A genome . | Uncultivable treponemes represent bacterial species and subspecies that are obligate pathogens of humans and animals causing diseases with distinct clinical manifestations ., Treponema pallidum subsp ., pallidum causes sexually transmitted syphilis , a multistage disease characterized in humans by localized , disseminated , and chronic forms of infection , whereas Treponema pallidum subsp ., pertenue ( agent of yaws ) and Treponema pallidum subsp ., endemicum ( agent of bejel ) cause milder , non-venereally transmitted diseases affecting skin , bones and joints ., The genetic basis of the pathogenesis and evolution of these microorganisms are still unknown ., In this study , a high quality whole genome sequence of the T . pallidum subsp ., endemicum Bosnia A strain was obtained using a combination of next-generation sequencing approaches and compared to the genomes of available uncultivable pathogenic treponemes ., Relative to all known genomes of Treponema pallidum subspecies , no major genome rearrangements were found in the Bosnia A . The Bosnia A strain clustered with other yaws-causing strains , while syphilis-causing strains clustered separately ., In general , the Bosnia A genome showed similar genetic characteristics to yaws treponemes but also contained several sequences thought to be unique to syphilis-causing strains ., This finding suggests a possible role of repeated horizontal gene transfer between treponemal subspecies in shaping the Bosnia A genome . | infectious diseases, medicine and health sciences, genetics, biology and life sciences, microbiology, molecular biology | null |
journal.pgen.1004450 | 2,014 | A Loss of Function Screen of Identified Genome-Wide Association Study Loci Reveals New Genes Controlling Hematopoiesis | Erythrocytes and platelets ( thrombocytes in zebrafish ) are the most abundant cells in blood ., In an individual , the number and volume of both erythrocytes and platelets are highly heritable and tightly regulated within narrow ranges , but there is a wide variation of these parameters in the population 1 , 2 ., The 80% heritability of blood cell indices provided the foundation for our recently completed GWAS meta-analysis in ∼68 , 000 healthy individuals for both cell types ., We identified 68 genetic loci that control the mass ( volume x count ) of platelets 3 and another 75 for red cell indices 4 ., About a quarter of the genes proximal to the platelet GWAS association single nucleotide polymorphisms ( SNPs ) encode well studied and generally pivotal regulators of hematopoiesis but the function of the remaining ones is unknown , demonstrating the power of GWAS to identify novel regulators of hematopoiesis ., We have recently reported functional validation of six genes in which the sentinel SNP was localized within a gene and silenced ak3 , rnf145 , arhgef3 , tpm1 , jmjd1c and ehd3 in zebrafish by MO injections ., Profound effects on thrombopoiesis were observed for all but ehd3 3 ., Furthermore , our detailed studies of the arhgef3 gene , which encodes one of the ∼70 Rho guanine nucleotide exchange factors , showed its important role in iron uptake and transferrin receptor internalisation in erythrocytes 5 ., Based on these preliminary data , we hypothesized that the majority of genes identified in our recent genomics efforts are important and rate-limiting regulators of hematopoiesis and therefore worthwhile of further investigation ., The zebrafish model has distinct advantages over other animal models for screening large numbers of genes ., Zebrafish development occurs rapidly over the course of a few days with thrombocytes , erythroid- and myeloid- blood cells being fully formed and functional by 3 days post fertilisation ( dpf ) ., External fertilisation and transparency of zebrafish embryos allow easy visualisation of early blood-related phenotypes giving them the advantage over mice , where development occurs in utero ., Importantly , transcriptional mechanisms and signalling pathways in hematopoiesis are well conserved between zebrafish and mammals 6 ., Herein , we performed a MO injection screen of 15 genes identified in GWAS for platelet size and number to uncover novel pathways essential in thrombopoiesis and hematopoiesis in general ., From this screen , we identified 12 new genes required for normal hematopoiesis and ordered them on a hematopoietic lineage tree based on their presumed function during hematopoiesis ., Further analysis of the hematopoietic lineage tree revealed a distinct pattern of gene distribution suggesting two main gene clusters ., One cluster of genes appears to work at the level of HSCs affecting all derived blood cell types and the second cluster appears to be limited to controlling the specification of the thrombocyte-erythroid progenitors ., Additionally , we show that one of the novel candidate genes , brd3a , is essential in differentiation of thrombocytes from HSCs but is dispensable for their maintenance ., To interrogate the large number of novel hematopoietic genes identified in the GWAS for platelet size and number we developed an in vivo functional genomics screen in zebrafish ( Figure S1 ) ., The first step was selection of the most suitable candidate genes and initially we selected a single gene , closest to the sentinel SNP , from each GWAS locus 3 ( Table S1 ) ., Distance from the nearest gene was calculated as the absolute distance between SNP and transcription start site of the gene or 3′ end of last exon 3 ( Table S1 ) ., We further eliminated genes with a known function in hematopoiesis and identified a putative zebrafish ortholog , with over 38% identity , on the protein level , with its human counterpart for 40 genes ., Finally , we excluded all genes that would require the use of more than two MOs , resulting in a list of 33 genes ., Nearly 80% of these genes had a sentinel SNPs localized within 10 kb ., In the last step , we selected 19 genes of which 16 had a sentinel SNP within 10 kb and three genes had a sentinel SNP >10 kb ., Five of the selected genes were duplicated in zebrafish , resulting in a total of 24 genes to be further investigated aiming to define their function in blood cell formation by a MO knockdown approach in zebrafish ., We designed splice-blocking MOs for each gene and validated their efficacy with RT-PCR and sequencing ., Of the 24 MOs tested five MOs had no effect on the target RNA and these MOs were excluded from further analysis ( Figure S2 ) ., We then assayed the remaining 19 functional MOs for their effect on overall development , morphology and hematopoiesis during the first 72 hours post fertilisation ( hpf ) and selected the optimal dose of MO to be injected ( Figure S3 ) ., Of note , based on the information available at the ZFIN at the time and our in situ hybridization data ( Figure S4 ) none of the selected genes had hematopoietic specific gene expression ., For all genes , except grtp1a , the optimal dose of MO was selected that resulted in a specific phenotype but without gross lethality or defects in body shape or size , vasculature , heart and circulation ., We found that grtp1a MO injected embryos died by 15 hpf even when injected with 0 . 8 ng of MO , thus we excluded grtp1a from further analysis ., Although morphological examination can detect defects with great sensitivity , some specific defects in hematopoiesis can be missed , and as a result information obtained from the initial analysis might be limited ., Thus , we carried out a second level of analysis by performing in situ hybridisation with several hematopoietic markers , specifically , c-myb , ae1 globin , mpeg and rag1 ., These were complemented with the use of the Tg ( cd41:EGFP ) line and two histochemical stains , o-Dianisidine and Sudan black ., The hematopoietic markers used for phenotyping were carefully chosen to distinguish between early and late stages of hematopoiesis as well as thrombocytes , erythrocytes , neutrophils , macrophages and lymphocytes ( Figures S5 ) ., We initiated screening by taking advantage of the Tg ( cd41:EGFP ) reporter line , which labels thrombocytes , to identify genes in zebrafish that , when disrupted , affect thrombocyte number ., We found that knock down of 15 of the 18 genes resulted in 30–95% reduction in thrombocyte number ( Figure 1 , Figure S6 ) ., Importantly , where the functional second MO was available , the observed phenotype was comparable to the one observed with the first MO ( Figure S7 , S8 , Table S2 ) ., Furthermore , concurrent knock down of p53 with gene specific MOs did not attenuate the thrombocyte phenotype induced by gene specific MOs , confirming that the observed decrease in the number of thrombocytes was not induced by p53 mediated off target effects of MO injection ( Figure S9 ) ., For one of the “phenotypic” genes , brf1b , we have obtained the mutant through ZMP ., To test if brf1b mutants have a decreased number of thrombocytes , the offspring were subjected to the “clotting time assay” at 5 dpf 7 ., Clotting time in brf1b mutant larvae was significantly longer than in the wild type fish , suggesting a defect in thrombopoiesis ( Figure S10 A ) ., To further confirm the specificity of the phenotype observed in MO injected embryos and in the absence of available mutants we performed the rescue of rcor1 MO injected embryos using full-length zebrafish rcor1 RNA ( Figure S10 B , C ) ., We focused our further analysis on the 15 “phenotypic” genes and their paralogs ., To maintain the population of differentiated blood cells within normal ranges , HSCs need to continuously maintain the balance between self-renewal and differentiation ., We reasoned that decreased number of thrombocytes in MO injected embryos could be the result of either reduced numbers of HSCs or altered HSC differentiation ., To assess which stage of hematopoiesis was affected in each MO injected embryo , we performed in situ hybridization at 3 dpf and looked for alterations in expression of definitive hematopoiesis marker c-myb ( Figure 2 ) ., Although more than half of the MOs tested had no effect on the number of HSCs , depletion of rcor1 resulted in an increased number of HSCs and depletion of kalrn ( 1 and 2 ) , mfn2 , pdia5 , psmd13 and wasplb resulted in decreased numbers of HSCs in caudal hematopoietic tissue ( CHT ) at 3 dpf ., This reduction in the number of HSCs was not evident at 30 hpf in kalrn1 , mfn2 , pdia5 , psmd13 and wasplb MO injected embryos ( Figure S11 ) suggesting that the number of HSCs at 3 dpf was most probably adversely affected by either their homing to or survival/proliferation in CHT ., However , kalrn2 and rcor1 depleted embryos had a marked decrease in the number of HSCs at 30 hpf , implying the important role of these genes in specification of HSCs in the aorta-gonad-mesonephros ( AGM ) ( Figure S11 ) ., Importantly , analysis of vascular development by injecting candidate gene MOs into Tg ( fli1:EGFP ) embryos , which express EGFP in endothelial cells , demonstrated no major abnormalities in vascular morphogenesis or remodeling that would preclude circulation , indicating that the hematopoietic defects were not secondary to a vascular phenotype ( Figure S12 ) ., Thus , our screen effectively defined a set of nine genes required for differentiation of HSCs to thrombocytes and possibly other blood lineages ., Hematopoiesis is often depicted by a hierarchical differentiation tree , with HSCs at the root and the mature blood cells as the branches ., One of the intermediate cellular states is the common myeloid progenitor ( CMP ) , which can proliferate and differentiate into megakaryocyte-erythrocyte progenitors ( MEP ) and granulocyte-monocyte ( GM ) progenitors , which further give rise to megakaryocytes , erythrocytes , granulocytes , monocytes and other cell types ., To investigate lineage-specific effects of the candidate genes , we assessed the status of definitive erythropoiesis in MO injected embryos at 4 dpf ., As αe1-globin RNA was reported to be expressed in definitive erythrocytes at 4 dpf 8 , a detailed analysis of the expression pattern of αe1-globin transcript was exploited to reveal the initiation of definitive erythropoiesis after gene silencing ., Profound effects on definitive erythropoiesis were observed for all but brd3a , brf1b , waspla and wdr66 ( Figure S13 ) ., However , silencing of brf1a and wasplb resulted in diminished definitive erythropoiesis , reflecting functional divergence of duplicated genes ( brf1 and waspl ) ( Figure S13 ) ., Furthermore , an extensive analysis of hemoglobin levels in primitive erythroid cells at 2 dpf showed that , with the exception of brd3a , kalrn2 and kif1b , primitive erythropoiesis was largely unaffected following MO knock down of candidate genes ( Figure S14 ) ., These results are consistent with the notion that the majority of candidate genes are dispensable for specification and differentiation of primitive erythrocytes and that fundamentally different molecular mechanisms regulate primitive and definitive erythropoiesis ., To establish the role of candidate genes in differentiation of the myeloid lineage , that is neutrophils and macrophages , we performed Sudan Black staining ( for neutrophils ) and in situ hybridization using mpeg riboprobe ( for macrophages ) in control and candidate gene depleted zebrafish embryos at 3 dpf ., Out of 15 genes tested , depletion of nine resulted in reduced numbers of Sudan Black positive cells and two ( kif1b , waspla ) had an effect on the number of macrophages ( Figures S15 , S16 , and S18 ) ., Reduction in the number of Sudan Black positive cells could reflect the absence of granules rather than neutrophils ., We have , therefore , performed in situ hybridization using mpx riboprobe for all genes for which the knockdown resulted in a decrease in the number of Sudan Black positive cells ., For all the tested genes the observed phenotype was comparable to the one we reported following Sudan Black staining ( Figure S17 ) ., Finally , we analyzed the impact of loss of candidate gene function on lymphoid development ., Differentiated thymic T-cells are exclusively derived from definitive HSCs and can be readily identified by rag1 expression when examined at 4 dpf ( Figure S19 ) ., Not surprisingly , a significant decrease in rag1 staining was mostly observed in the same set of genes in which we observed a decrease in c-myb staining ( a marker for HSCs ) , namely kalrn1 , kalrn2 , mfn2 , pdia5 and psmd13 ., In addition , injection of akap10 MO and rcor1 MO , which had no negative impact on the number of c-myb positive cells , resulted in a significant decrease in the number of T lymphocytes ., The large number of genes analyzed and the resultant volume of data acquired can present challenge in understanding and interpreting the results ., Hence , we used the information gained from the initial MO knockdown screen to generate a heat-map of phenotype profiles ( Figure, 3 ) and cluster genes with similar phenotypic profiles ., We then hierarchically positioned candidate genes on the hematopoietic lineage tree and assigned each of them a potential role during hematopoietic differentiation ( Figure 4 ) ., Our analysis of the hematopoietic lineage tree revealed a distinct pattern of gene distribution , suggesting two main gene clusters ., The first cluster represents a set of genes , namely kalrn1 and -2 , mfn2 , pdia5 , psmd13 , rcor1 and wasplb , which upon depletion affect the number of HSCs ., The second major cluster represents a set of genes , namely akap10 , brf1a , kif1b , satb1 and wasplb , which appear to be essential further down the hematopoietic tree and affect differentiation of both definitive erythrocytes and thrombocytes ., These genes have presumed role in HSC fate decisions prior to specification of the thrombocyte and erythrocyte progenitors ., Although the frequency of blood defects observed in our screen was high , the screen was not as well suited for the identification of knockdown phenotypes that result in subtle differences in myeloid lineage cell production or skewing of myeloid lineage differentiation ., This is mainly because changes in the number of neutrophils and macrophages arising from HSCs may be undetectable using markers and the developmental time point outlined here ., Previous studies have shown that erythroid-myeloid progenitor cells ( EMPs ) are capable of generating both macrophages and neutrophils and that these blood cells appear in mib zebrafish despite the absence of HSCs 9 ., However , even with these caveats , we believe the screening procedure used has proven effective for extracting functional information from a GWAS dataset in a medium-throughput manner ., To gain an additional insight into mechanisms by which these newly discovered genes affect thrombopoiesis , we performed a more extensive analysis of the function of brd3a in hematopoiesis ., The bromodomain and extra terminal domain ( BET ) family of proteins , including BRD2 , BRD3 , and BRD4 , are evolutionally conserved and play a key role in many cellular processes by controlling the assembly of histone acetylation-dependent chromatin complexes 10 ., To further confirm that the defects observed in the brd3a depleted embryos resulted from loss of brd3a , in vitro–transcribed RNA encoding human BRD3 ( hBRD3 ) was injected into 1-cell stage embryos ., Live confocal imaging of zebrafish embryos injected with hBRD3-GFP confirmed that hBRD3 binds to mitotic chromosomes ( Figure 5A ) , a feature previously reported for BET family proteins i . e . BRD2 , BRD3 and BRD4 11–13 ., Expression of hBRD3 in brd3a MO injected embryos resulted in partial but statistically significant rescue of the number of thrombocytes demonstrating that brd3a MO used in this study exerted a specific effect ( Figure 5B , C ) ., Morpholinos do not allow gene-specific perturbation to be carried out with temporal resolution , which is a disadvantage when dissecting the precise role of a selected gene in hematopoiesis ., A number of studies reported that compounds targeting BET proteins might be used to manipulate hematopoietic development for exploratory or therapeutic purposes 14–16 ., The BET family inhibitor , thieno-triazolo-1 , 4-diazepine ( ( + ) -JQ1 in short ) is a potent , highly specific inhibitor which displaces BET proteins from chromatin by competitively binding to the acetyl-lysine recognition pocket of BET bromodomains 17 , 18 ., Thus , we evaluated the pharmacological impact of ( + ) -JQ1 on zebrafish development and thrombopoiesis ., Exposure of zebrafish embryos to ( + ) -JQ1 disrupted the chromatin occupancy of hBRD3-GFP confirming the efficacy of the inhibitor ( Figure S20 A–C ) ., We next incubated embryos from 6 hpf in various concentrations of ( + ) -JQ1 and ( − ) -JQ1 ( stereoisomer which has no appreciable affinity to BET bromodomains ) 17 as a control ( Figure S21 ) ., Exposure of zebrafish embryos to 1 µg/ml ( + ) -JQ1 resulted in complete mortality by 24 hpf whereas the ( − ) -JQ1 enantiomer showed no observable effect on embryo development ( Figure S21 ) ., This early embryonic death of zebrafish embryos was not surprising considering that knockout of Brd2 or Brd4 in mice results in embryonic lethality , indicating an important role of these two proteins in embryonic development 19 , 20 ., When treated , however , with ( + ) -JQ1 from 24 hpf , embryos exhibited overall normal development even at the higher concentration ( 1 µg/ml ) of ( + ) -JQ1 ( Figure S21 ) ., Although morphologically normal , thrombopoiesis was completely abolished in these embryos ( Figure 6 A ) ., Interestingly , the decrease in the number of thrombocytes appeared more prominent in the presence of ( + ) -JQ1 inhibitor compared to brd3a MO knock down ., This opened the possibility that other members of the BET family might be contributing to the observed phenotype ., To investigate this further , we performed MO knock down of zebrafish brd2a , brd2b and brd4 and assessed the number of thrombocytes at 3 dpf ., Single MO knock down of all three genes resulted in a severe decrease in the number of thrombocytes ., These data strongly suggested that , indeed , other members of BET family of proteins ( i . e . brd2 and brd4 ) play an important role in thrombopoiesis ( Figure S22 A–C ) ., Both MO knock down and treatment with ( + ) -JQ1 inhibitor from 24 hpf resulted in a severe reduction in the number of thrombocytes at 3 dpf , suggesting an essential role for brd3a in the differentiation of thrombocytes as opposed to a requirement for their maintenance and survival ., To address this question we incubated Tg ( cd41:EGFP ) embryos with ( + ) -JQ1 inhibitor starting from 3 dpf , when there is already a considerable number of thrombocytes in CHT , and assessed their number 24 hours later , at 4 dpf ., We found that in untreated and ( − ) -JQ1 treated embryos the number of thrombocytes markedly increased between 3 and 4 dpf ., However , in ( + ) -JQ1 treated embryos we observed no change in the number of thrombocytes ( Figure 6C ) ., Moreover we found that ( + ) -JQ1 had no adverse effect on the number of HSCs during this 24 h period of treatment ( Figure 6D ) ., Taken together this strongly suggests that brd3a is important in differentiation of thrombocytes from HSCs , however , once differentiated , brd3a was dispensable for their maintenance and survival ., GWAS meta-analysis of platelet size and number has been successful in identifying SNPs associated with the mass ( volume x count ) of platelets ., In contrast with the results of GWAS in common diseases , more than 80% of SNPs associated with hematological traits are localized within 10 Kb of genes providing a sound argument to infer biologically relevant candidate genes 3 , 4 ., Canonical pathway analyses detected a highly significant over-representation of “core genes” ( the sentinel SNP is within the gene or within 10 kb from the gene ) in relevant biological functions such as hematological disease , cancer and cell cycle 3 ., However , three quarters of regions proximal to the platelet GWAS SNPs harbor unfamiliar genes or known genes not previously implicated in hematopoiesis that merit extensive follow-up analysis ., Therefore , this study was set up to address the need for a medium-throughput method in zebrafish to dissect the functional roles of these assumed novel regulators of hematopoiesis ., In total , our screen identified 15 genes ( corresponding to 12 human genes ) required for distinct stages of specification or differentiation of HSCs in zebrafish ., A detailed review of the content of databases and literature revealed limited knowledge about the functional role of Satb1 , Rcor1 and Brd3 21–24 in hematopoiesis and for the remaining nine genes our work represents the first study on their putative role in hematopoiesis ., Importantly , our results are well in line with some of the findings reported by others ., One example is RCOR1 - lineage-restricted deployment of RCOR1 and LSD1 cofactors , through interaction with Gfi proteins , controls hematopoietic differentiation 23 ., Knock down of rcor1 in zebrafish resulted in completely blocked differentiation of erythroid , thrombocytic , myeloid and lymphoid lineages ., These findings strongly support the hypothesis that the published platelet GWAS 3 enriched for functional regulators of the hematopoiesis and further support previous assumptions that a large proportion of the genes uncovered by the aforementioned GWAS also have a conserved role in zebrafish ., In this study , we followed a two step screening approach: in the first instance , we used the Tg ( cd41:EGFP ) line in conjunction with a panel of hematopoietic in situ hybridization probes and histochemical staining to create a heat map with distinct “phenotype signatures” of each gene knock-down ., We then positioned the candidate genes on the hematopoietic cell lineage tree and assigned them a potential role in hematopoietic differentiation ., Interestingly , our screen revealed that , although initially selected based on their effect on the platelet size and/or number , none of the candidates exerted a thrombocyte specific effect ., These results should be interpreted within the context of several major differences between the effect of the SNPs and whole embryo MO knock down on hematological traits ., First , the majority of associated SNPs identified in platelet GWAS are in non-transcribed regions and it is likely that the underlying mechanism linking them to the phenotype is regulatory ., Thus , the functional effects of SNPs are subtler compared to the knock down of transcripts achieved by MOs in our screening ., Secondly , although GWAS provided a list of SNPs associated with the platelet size and number , there is no evidence about the biological processes that link the associated SNP to the phenotype ., Indeed , it has been shown that in most cases the reported SNP is not the functional SNP itself but is in linkage disequilibrium with the SNP overlapping a functional region 25 ., Experimental evidence shows that open chromatin profiles of megakaryocytes and erythroblasts differ and thus cell type-restricted regions of open chromatin could influence the penetrability of the functional SNP 26 , 27 in a lineage specific manner ., In contrast , MO knock down in zebrafish is not spatially restricted and thus offers the opportunity to determine the functional role of candidate genes in all blood lineages ., To further verify the hematopoietic role of genes identified in GWAS , we performed a more extensive evaluation of the effect of brd3a on thrombopoiesis ., It has been shown that BRD3 interacts with acetylated GATA1 and stabilizes its chromatin occupancy 21 ., A pharmacologic compound , JQ1 , that occupies the acetyl-lysine binding pockets of Brd3 bromodomains disrupts the BRD3-GATA1 interaction , diminishes the chromatin occupancy of both proteins , and inhibits erythroid maturation 21 ., Although GATA1 and BRD3 co-occupancy on GATA1 target genes was also observed in a megakaryocytic cell line 21 , the biological relevance of this binding was never confirmed ., Here we report on the important role of brd3a in thrombopoiesis ., Indeed , knock-down of brd3a with two independent MOs as well as treatment of zebrafish embryos with the JQ1 inhibitor starting from 24 hpf severely reduced the number of thrombocytes in 3 days old embryos ., Interestingly , incubation of embryos with JQ1 inhibitor between 3- and 4 dpf , that is after the onset of thrombopoiesis , did not have any effect on the already differentiated thrombocytes ., However , the number of thrombocytes failed to increase compared to control embryos within this 24-hour period ., These results strongly support the idea that brd3a is critical for establishing but not maintaining thrombopoietic compartment ., Some previous studies suggest that BRD3 , as well as some other mitotically retained factors , functions as a molecular “bookmark” by enabling post-mitotic transcription re-initiation of its target genes 13 , 28 ., It is plausible to assume that a similar mechanism is employed during thrombopoiesis ., In that scenario , retention of Brd3 on chromatin during mitosis of thrombocyte precursors or erythroid-thrombocyte progenitor cells would contribute to the maintenance of transcription patterns necessary for establishment of thrombocyte identity ., However , further work will be necessary to identify the precise molecular mechanisms by which brd3a exerts its effect on thrombopoiesis ., Taken together , our study provides a paradigm of the usefulness of zebrafish for efficient translation of GWAS findings into relevant biological information in an objective and unbiased manner ., GWAS have mapped many novel , convincingly associated loci in the proximity of genes where functional significance is expected ., So far , functional validation of such genes has remained confined to single gene approaches ., Here we utilized the powerful genetics and translucency of zebrafish larvae to undertake a medium-throughput screen of genes implicated in human hematopoietic variation ., The results of this screen will help us to tentatively place novel genes in molecular pathways and thus close the ever-increasing knowledge gap on the biological function of gene candidates identified by genomic technologies ., The maintenance , embryo collection and staging of the wild type ( Tubingen Long Fin ) and transgenic zebrafish lines ( Tg ( cd41:GFP ) , Tg ( fli1:GFP ) , Tg ( c-myb:EGFP ) were performed in accordance with EU regulations on laboratory animals , as previously described 29 , 30 ., Morpholinos ( GeneTools , LLC ) were re-suspended in sterile water and diluted to chosen concentration ., Approximately 1 nl was injected into embryos at 1- to 2-cell stage ., MOs used are summarized in Table S2 ., Plasmid with full-length human hBRD3 cDNA was purchased from Source Bioscience ( Nottingham , UK ) ., hBRD3 was cloned into pCS2 expression vector using gene-specific primers: AATTACATCGATACCATGTCCACCGCCACGACAGT ( forward ) and CCCGAGTCTAGACTATTCTGAGTCACTGCTGTCAGA ( reverse ) and AAATTAGAATTCACCATGTCCACCGCCACGACAGT ( forward ) and ATGTTAACCGGTAGTTCTGAGTCACTGCTGTCAGA ( reverse ) for cloning into the pCS2-EGFP vector ., Restriction enzyme sites ( ClaI/XbaI and EcoRI/AgeI respectively ) used for cloning are underlined ., Zebrafish full-length rcor1 cDNA was cloned into pCS2 vector using gene-specific primers: GTTATAGAATTCATGCCCGCAATGTTAGAGAAG ( forward ) and AGGCGCCTCGAGTCAGGAAACCGAAGGGTTCTG ( reverse ) ., Restriction enzyme sites ( EcoRI and XhoI , respectively ) are underlined ., hBRD3 , hBRD3-GFP and rcor1 mRNA was synthesized with mMESSAGE mMACHINE kit ( Ambion ) , according to the manufacturers protocol ., In the rescue experiment , 100 pg of hBRD3 mRNA or 125 pg rcor1 RNA was injected into the one cell stage control and MO-injected Tg ( cd41:EGFP ) embryos ., In the hBRD3 localization experiment , 300 pg of hBRD3-GFP mRNA was injected into Tubingen Long Fin embryos at 1-cell stage ., In order to verify the effectiveness of MOs in affecting their target transcripts , RT-PCR was performed ., RNA was subjected to reverse transcription using Superscript II Reverse Transcriptase ( Invitrogen ) ., PCR was performed using gene-specific primers ( listed in Table S2 ) and KOD Hot Start DNA Polymerase ( Novagen ) ., In situ hybridization was performed with riboprobes specific for c-myb , αe1-globin , mpx , mpeg and rag1 as previously described 31 , as well as for brd3a , brf1b , kalrn1 , waspla and wdr66 ., Primers used for PCR amplification of candidate genes for probe synthesis are listed in Table S3 ., Photomicrographs were taken with a Zeiss camera AxioCam HRC attached to a LeicaMZ16 FA dissecting microscope ( Leica Microsystems , Germany ) using the AxioVision software ., O-dianisidine staining was performed as previously described 32 ., Sudan Black staining was performed as previously described 33 ., Clotting time assay was performed as previously described 7 ., In short , 5 dpf larvae were anaesthetized in 0 . 02% tricaine solution in embryonic water and transferred onto a Petri dish in a small drop of liquid ., Caudal veins of the larvae were wounded with the tip of a Microlance needle ( 0 . 4 mm×13 mm , Becton Dickinson ) in the anal area ., For each larva the time passing between inflicting the wound and the stop of bleeding was recorded ., Genomic DNA was isolated from 5 dpf larvae , which were individually loaded into wells of a 96-well plate ., Fish were incubated in 25 µl of lysis buffer ( 25 mM NaOH with 0 . 2 mM EDTA ) at 95°C for 30 min ., Afterwards 25 µl of neutralization buffer ( 40 mM Tris-HCl ) was added ., Genotyping was carried out using the KASP genotyping assays ( KBioscience ) ., Each reaction consisted of 4 µl of genomic DNA and 5 µl of PCR mix , according to the manufacturers protocol ( KBioscience ) ., PCR products were analyzed using PHERAstar plus ( BMGlabtech ) and KlusterCaller software ( KBioscience ) ., Images were captured with the use of a Leica TCS SP5 confocal microscope with Leica LAS AF software ( Leica Microsystems ) , using a 40× immersion lens or with Axio Zoom . V16 fluorescent microscope with AxioCam MRm camera using 260× magnification ., Selective inhibitor of human BET family of bromodomain-containing proteins , thieno-triazolo-1 , 4-diazepine , named JQ1 , was kindly provided by Dr Chas Bountra , Structural Genomics Consortium , University of Oxford , Oxford , UK ., Both the active inhibitor , ( + ) -JQ1 , and its inactive stereoisomer , ( − ) -JQ1 , were dissolved in dimethylsulfoxide ( DMSO ) to 10 mg/ml and stored in aliquots at −20°C ., For zebrafish embryo treatment , ( + ) -JQ1 and ( − ) -JQ1 were diluted in egg water to desired concentration and added to the embryos at ∼6 hpf , 24 hpf or 3 dpf and afterwards replaced daily ., Control embryos were incubated in the equal concentration of DMSO in egg water as inhibitor-treated embryos . | Introduction, Results, Discussion, Materials and Methods | The formation of mature cells by blood stem cells is very well understood at the cellular level and we know many of the key transcription factors that control fate decisions ., However , many upstream signalling and downstream effector processes are only partially understood ., Genome wide association studies ( GWAS ) have been particularly useful in providing new directions to dissect these pathways ., A GWAS meta-analysis identified 68 genetic loci controlling platelet size and number ., Only a quarter of those genes , however , are known regulators of hematopoiesis ., To determine function of the remaining genes we performed a medium-throughput genetic screen in zebrafish using antisense morpholino oligonucleotides ( MOs ) to knock down protein expression , followed by histological analysis of selected genes using a wide panel of different hematopoietic markers ., The information generated by the initial knockdown was used to profile phenotypes and to position candidate genes hierarchically in hematopoiesis ., Further analysis of brd3a revealed its essential role in differentiation but not maintenance and survival of thrombocytes ., Using the from-GWAS-to-function strategy we have not only identified a series of genes that represent novel regulators of thrombopoiesis and hematopoiesis , but this work also represents , to our knowledge , the first example of a functional genetic screening strategy that is a critical step toward obtaining biologically relevant functional data from GWA study for blood cell traits . | In this manuscript we report on a follow-up study of the GWAS loci associated with the platelet size and number ., A GWAS meta-analysis identified 68 genetic loci controlling platelet size and number ., Only a quarter of those genes , however , are known regulators of hematopoiesis ., To determine function of the remaining genes we performed a medium-throughput genetic screen in zebrafish using morpholinos ( MOs ) to knock down selected candidate genes ., Here , we report on two major findings ., First we identified 15 genes ( corresponding to 12 human genes ) required for distinct stages of specification or differentiation of HSCs in zebrafish ., A detailed review of databases and literature revealed limited knowledge about the functional role of Satb1 , Rcor1 and Brd3 in hematopoiesis and for the remaining nine genes our work represents the first study on their putative role in hematopoiesis ., And secondly , we demonstrate that brd3a is critical for establishing , but not maintaining thrombopoietic compartment ., Importantly , our study introduces zebrafish as a model system for functional follow-up of GWAS loci and generates a valuable resource for prioritization of platelet size and number associated genes for future in-depth mechanistic analyses ., Following this route of investigation new regulatory molecules of hematopoiesis will be added to critical pathways . | developmental biology, medicine and health sciences, body fluids, blood, anatomy, platelets, biology and life sciences, hematology | null |
journal.pntd.0001539 | 2,012 | Clear Genetic Distinctiveness between Human- and Pig-Derived Trichuris Based on Analyses of Mitochondrial Datasets | Soil-transmitted helminths ( =\u200ageohelminths ) , including whipworm , are responsible for neglected tropical diseases ( NTDs ) of humans in developing countries 1–3 ., Trichuris trichiura infects ∼600 million people worldwide ., This parasite is transmitted directly via a direct , faecal-oral route ., The thick-shelled ( infective ) eggs are ingested and then hatch , following gastric passage , in the small intestine ., First-stage larvae ( L1s ) are released and migrate to the large intestine ( caecum and colon ) , where they develop , following multiple moults , into adults ( ∼30–50 mm in length ) ., The worms burrow their thin , thread-like anterior end into the mucosal lining of the large intestinal wall , feed on tissue fluids , mature and produce eggs ., In the large intestines , large numbers of worms cause disease ( =\u200atrichuriasis ) , which is usually associated with entero-typhlocolitis and clinical signs , such as dysentery , bloody diarrhoea and/or rectal prolapse , in people with a high intensity of infection ., Children ( ∼5–15 years of age ) often harbour the largest numbers of worms 2 ., Whipworms also infect other animal hosts , including non-human primates , pigs and dogs , and can cause clinical disease similar to trichuriasis of humans 4–6 ., Based on current knowledge , Trichuris species are considered to specifically infect a particular host species or a group of related hosts ., Trichuris species are usually identified based on host origin and the morphological features of the adult worm ( spicule and pericloacal papillae ) 7 , 8 ., However , it is not always possible to unequivocally identify and differentiate Trichuris species based on the morphology of adult worms alone ., Importantly , T . trichuria cannot be unequivocally differentiated morphologically from T . suis or Trichuris from some other animals , such as non-human primates 7 ., Over the years , there has been considerable discussion as to whether T . trichuira and T . suis are the same or distinct species 9–13 , and whether humans can become infected with T . suis , and pigs with T . trichiura in endemic countries in which both host species live in close association ., Although the authors of a recent molecular study suggested that T . suis is a separate species from T . trichiura 14 , only a small number of specimens from one country ( Spain ) was used in this study , and amplicons ( from the first and second internal transcribed spacers , ITS-1 and ITS-2 , of nuclear ribosomal DNA ) were subjected to cloning prior to sequencing , which has significant potential to lead to artefacts 15 , 16 ., Therefore , the findings from this study 14 need to be interpreted with some caution at this stage ., Moreover , internal transcribed spacers ( ITS ) of nuclear ribosomal DNA might not be suited as specific markers for enoplid nematodes , because of sequence polymorphism ( heterogeneity ) that occurs within species ( or individuals ) 14 , 17 ., Given this heterogeneity in nuclear rDNA , barcoding from whole mitochondrial ( mt ) genomes ( haploid ) has major advantages , particularly when concatenated protein sequences derived from all coding genes are used as markers in comparative , phylogenetic-based analyses 18–25 ., Therefore , in the present study , we, ( i ) characterised the mt genomes of human-derived Trichuris and pig-derived Trichuris ,, ( ii ) compared these genomes and, ( iii ) then tested the hypothesis that human-Trichuris and pig-Trichuris are genetically distinct in a phylogenetic analysis of sequence data sets representing both genomes and those from selected nematodes for comparative purposes ., This study was approved by the Animal Ethics Committee of the Lanzhou Veterinary Research Institute , Chinese Academy of Agricultural Sciences ., For the collection of Trichuris from humans , the subjects provided informed , written consent ., All pigs , from which Trichuris specimens were collected , were handled in accordance with good animal practices required by the Animal Ethics Procedures and Guidelines of the Peoples Republic of China ., Adult specimens of Trichuris were collected from the caecum of a human patient during surgery in Zhanjiang Peoples Hospital in Zhanjiang , Guangdong Province , China ., Adult specimens of Trichuris were also collected from the caecum from a pig slaughtered in an abattoir in Zhanjiang in the same province ., Adult worms from each host were washed separately in physiological saline , identified morphologically 9 , 26 , fixed in 70% ( v/v ) ethanol and stored at −20°C until use ., Total genomic DNA was isolated separately from two individual worms ( coded Ttr2 and TsCS1 for human-Trichuris and pig-Trichuris , respectively ) using an established method 27 ., The region spanning ITS-1 , the 5 . 8S gene and ITS-2 was amplified from each of these individuals by PCR using previously reported primers 14 and sequenced directly ., The ITS-1 sequence of the human-Trichuris sample had 99 . 3% similarity with that of T . trichiura from human in Thailand ( GenBank accession no . GQ352554 ) ., The ITS-1 and ITS-2 sequences of the pig-Trichuris sample had 98 . 6% and 98 . 5% similarity with that of T . suis from pigs in Spain ( GenBank accession nos . AJ781762 and AJ249966 , respectively ) 14 ., To obtain some mt gene sequence data for primer design , we amplified regions ( 400–500 bp ) of the cox1 and nad1 genes by using ( relatively ) conserved primers JB3/JB4 . 5 and JB11/JB12 , respectively 28 , and of nad4 and rrnL genes using primers designed in this study ( Table, 1 ) by PCR ., The amplicons were sequenced from both directions , using BigDye terminator v3 . 1 , ABI PRISM 3730 ., We then designed primers ( see Table, 2 ) to regions within cox1 , nad1 , nad4 and rrnL and amplified from total genomic DNA ( from an individual worm ) the entire mt genome in three ( for human-Trichuris ) or four ( for pig-Trichuris ) overlapping fragments ( of ∼2–4 kb each ) between nad1 and nad4 , nad4 and rrnL , and rrnL and cox1 , cox1 and nad1 ., The cycling conditions used were 92°C for 2 min ( initial denaturation ) , then 92°C for 10 s ( denaturation ) , 50°C for 30 s ( annealing ) , and 60–68°C for 10 min ( extension ) for 10 cycles , followed by 92°C for 10 s , 50°C for 30 s , and 60–68°C for 10 min for 20 cycles , with a cycle elongation of 10 s for each cycle and a final extension at 60–68°C for 7 min ., Each amplicon , which represented a single band in a 0 . 8% ( w/v ) agarose gel , following electrophoresis and ethidium-bromide staining , was column-purified and then sequenced using a primer walking strategy 29 ., Sequences were assembled manually and aligned against the complete mt genome sequences of other nematodes ( available publicly ) using the computer program Clustal X 1 . 83 30 to infer gene boundaries ., The open-reading frames ( ORFs ) and codon usages of protein-coding genes were predicted using the program MacVector v . 4 . 1 . 4 ( Kodak ) , and subsequently compared with that of Trichinella spiralis 31 ., Translation initiation and translation termination codons were identified based on comparison with those reported previously 31 ., Codon usages were examined based on the relationships between the nucleotide composition of codon families and amino acid occurrence , for which codons are partitioned into AT rich codons , GC-rich codons and unbiased codons ., The secondary structures of 22 tRNA genes were predicted using tRNAscan-SE 32 and/or manual adjustment 33 ., Amino acid sequences inferred from the 12 protein-coding genes ( i . e . not atp-8 ) common among all of the nematodes included here were concatenated into a single alignment , and then aligned with those of 9 other enoplid nematodes ( Trichinella spiralis , GenBank accession number NC_002681; Xiphinema americanum , NC_005928; Hexamermis agrotis , NC_008828; Agamermis sp . , NC_008231; Romanomermis culicivorax , NC_008640; Romanomermis iyengari , NC_008693; Romanomermis nielseni , NC_008692; Strelkovimermis spiculatus , NC_008047; Thaumamermis cosgrovei , NC_008046 ) , using the chromadorean nematode , Brugia malayi ( NC_004298 ) as the outgroup ., Any regions of ambiguous alignment were excluded using Gblocks ( http://molevol . cmima . csic . es/castresana/Gblocks_server . html; Talavera and Castresana 2007 ) using stringent selection criteria ( do not allow many contiguous nonconserved positions ) ., Phylogenetic analyses were conducted using three methods: Bayesian inference ( BI ) analysis was conducted with four independent Markov chain runs for 1 , 000 , 000 metropolis-coupled MCMC generations , sampling a tree every 100th generation in MrBayes 3 . 1 . 1 34; the first 2 , 500 trees represented burn-in , and the remaining trees were used to calculate Bayesian posterior probabilities ( pp ) ., Maximum likelihood ( ML ) analyses were performed using PhyML 3 . 0 35 , and the ( mtREV for amino acid sequences and GTR for rrnL nucleotide sequences ) models were determined based on the Akaike information criterion ( AIC ) ., Bootstrap support was calculated using 100 bootstrap replicates ., Maximum parsimony ( MP ) analysis was conducted using PAUP 4 . 0 Beta 10 36 , with indels treated as missing character states; 1 , 000 random additional searches were performed using TBR ., Bootstrap support was calculated using 1 , 000 bootstrap replicates , and 100 random taxon additions in PAUP ., Phylograms were drawn using the program Tree View v . 1 . 65 37 ., Two primers , rrnLF ( 5′-TAAATGGCCGTCGTAACGTGACTGT-3′ ) and rrnLR ( 5′- AAAGAGAATCCATTCTATCTCGCAACG-3′ ) , were employed for PCR amplification and subsequent sequencing of a portion ( 471 bp for human-Trichuris and 482 bp for pig-Trichuris ) of the large subunit of mt ribosomal RNA ( rrnL ) from multiple individuals of human- and pig-derived Trichuris ( Table 3 ) ., The rrnL sequence from T . spiralis ( accession number NC_002681 ) 31 was used as the outgroup in phylogenetic analyses , because this morphologically distinct species is related to Trichuris 38 ., All rrnL sequences were aligned using Clustal X , and the alignment was modified manually , based on the predicted secondary structure of the rrnL for Trichuris 31 , and then subjected to phylogenetic analysis using the same methods as described above ., The complete mt genome sequences were 14 , 046 nt ( human-Trichuris ) and 14 , 436 nt ( pig-Trichuris ) in length , respectively ( GenBank accession numbers GU385218 and GU070737 ) ., Each mt genome contains 13 protein-coding genes ( cox1-3 , nad1-6 , nad4L , cytb , atp6 and atp8 ) , 22 transfer RNA genes and two ribosomal RNA genes ( rrnS and rrnL ) ( Table 4 ) ., The genes nad6 and cox3 overlapped by 25 bp ( human-Trichuris ) ; nad5 overlaps by 7 and 1 bp with tRNA-His , tRNA-Ser ( AGN ) by 8 and 3 bp with rrnS , tRNA-Asp by 13 and 4 bp with atp8 for human-Trichuris and pig-Trichuris , respectively ., The atp8 gene is encoded ( Fig . 1 ) , as is typical for adenophorean nematodes 31 ., The protein-coding genes are transcribed in different directions , as described for T . spiralis and X . americanum 31 , 39 ., Except for four protein-coding genes ( nad2 , nad5 , nad4 and nad4L ) and six tRNA genes ( tRNA-Arg , tRNA-His , tRNA-Met , tRNA-Phe , tRNA-Pro and tRNA-Thr ) encoded on the L-strand , all other genes were encoded on the H-strand ., The AT-rich regions are located between nad1 and tRNA-Lys , and nad3 and tRNA-Ser ( UCN ) , which differs from those of secernentean nematodes 19 , 33 ., The nucleotide composition of the entire mt genome is biased toward A and T , with T being the most favoured nucleotide and G being least favoured , which is consistent with mt genomes of some other nematodes for which mt genomic data are available 18 , 19 , 21 , 31 , 33 ., The overall A+T content is 68 . 1% for human-Trichuris ( 33 . 6% A , 34 . 5% T; 15 . 0% G and 16 . 9% C ) and 71 . 5% for pig-Trichuris ( 35 . 6% A , 35 . 9% T; 13 . 4% G and 15 . 1% C ) ., Protein-coding genes were annotated by aligning sequences and identifying translation initiation and termination codons by comparison with inference sequences for other nematodes ., For both human-Trichuris and pig-Trichuris , the lengths of protein-coding genes were in the following order: nad5 ( 1548–1557 bp ) >cox1>nad4>cytb>nad1>nad2>atp6>cox3>cox2>nad6>nad3>nad4L>atp8 ( 165–171 bp ) ( Table 4 ) ., The longest gene is nad5 , and the lengths of the nad1 and nad3 genes are the same for human-Trichuris and pig-Trichuris ( Table 5 ) ., The inferred nucleotide and amino acid sequences of each of the 13 mt proteins of human-Trichuris were compared with pig-Trichuris ., For individual genes , the nucleotide and amino acid sequence differences between human-Trichuris and pig-Trichuris vary from 25 . 4 to 37 . 4% and 13 . 6 to 62 . 5% , respectively ( Table 5 ) ., A total of 3559 and 3562 amino acids are encoded in the mt genome of human-Trichuris and pig-Trichuris , respectively ., As the mt genomes of nematodes can contain non-standard initiation codons 18 , the identification of initiation codons can sometimes be challenging ., For human-Trichuris , five genes ( cox1 , cox2 , cox3 , cytb and nad4 ) start with ATG and eight genes ( nad1 , nad2 , nad3 , nad5 , nad6 , nad4L , atp6 and atp8 ) use ATA ., All genes have complete termination codons , 11 genes ( cox1 , cox2 , cox3 , nad1 , nad2 , nad3 , nad4 , nad4L , nad5 , nad6 and atp6 ) use TAA and two genes ( atp8 and cytb ) use TAG as a termination codon , respectively ., For pig-Trichuris , six genes ( nad2 , nad3 , nad4 nad4L nad5 and atp6 ) start with ATA , four genes ( cox1 , cox2 , cox3 and cytb ) with ATG , and one gene ( atp8 ) with TTG ., All protein-coding genes have complete termination codon; seven genes ( cox2 , cox3 , nad3 , nad4 , nad6 , atp6 and atp8 ) stop with TAA and six genes ( cox1 , nad1 , nad2 , nad5 , nad4L and cytb ) use TAG ., No abbreviated stop codons , such as TA or T , were detected ., Such codons are known to occur in the mt genomes of other nematodes , such as Strongyloides stercoralis ( cytb , nad4 , nad1 and atp6 ) and T . spiralis ( cytb and nad4 ) and Caenorhabditis elegans ( nad1 and nad3 ) 19 , 31 , 40 ., The atp8 gene was inferred by comparison with the homologous gene of T . spiralis ., The ORF of this gene was located between genes tRNA-Asp and nad3 and inferred by the presence of a methionine for human-Trichuris and a leucine for pig-Trichuris ., Twenty-two tRNA genes were predicted from the mt genomes of human-Trichuris and pig-Trichuris and varied from 50 to 67 nt in length ., Most of the tRNA genes are smaller than the corresponding genes in the mt genomes of other nematodes due to a reduced TΨC stem-loop region ( TV-replacement loop ) or DHU stem-loop region 39 ., Most of the tRNA gene sequences can be folded into conventional secondary four-arm cloverleaf structures ., In these tRNA , there is a strict conservation of the sizes of the amino acid acceptor stem ( 11–15 bp ) and the anticodon loop ( 7 bp ) ., Their D-loops consist of 5–9 bp ., The two tRNA-Ser each contain the TΨC arm and loop , but lack the DHU arm and loop ., The two ribosomal RNA genes ( rrnL and rrnS ) of human-Trichuris and pig-Trichuris were inferred based on comparisons with sequences from T . spiralis; rrnL is located between tRNA-Val and atp6 , and rrnS is located between tRNA-Ser ( AGN ) and tRNA-Val ., The length of rrnL is 1011 bp for both human-Trichuris and pig-Trichuris ., The lengths of the rrnS genes are 698 bp for human-Trichuris and 712 bp for pig-Trichuris ., The A+T contents of rrnL for human-Trichuris and pig-Trichuris are 72 . 5% and 76 . 4% , respectively ., The A+T contents of rrnS for human-Trichuris and pig-Trichuris are 69 . 9% and 75 . 4% , respectively ., Two AT-rich non-coding regions ( NCRs ) were inferred in the mt genomes of both human-Trichuris and pig-Trichuris ., For these genomes , the long NCR ( designated NCR-L; 162 bp and 144 bp in length , respectively ) is located between the nad1 and tRNA-Lys ( Fig . 1 ) , has an A+T content of 71–72% ., This overall A+T content is lower than those reported for nematodes ( 77 . 9–93 . 1% ) studied to date 18 , 19 , 21 , 33 ., In this NCR , there are also 26 nt ( human-Trichuris ) and 17 nt ( pig-Trichuris ) AT dinucleotide repeats ., Similar repeats have been detected in this region in C . elegans and A . suum 40 ., For both human-Trichuris and pig-Trichuris , the short NCR ( NCR-S; 93 bp and 117 bp in length ) is located between genes nad3 and tRNA-Ser ( UCN ) ( Fig . 1 ) , with an A+T content of 65 . 6% and 84 . 6% , respectively ., This region contains dinucleotide AT26 repeats and might form a hairpin loop structure ( cf . AAAAAAAATTTTTTTTTT ) ., Although nothing is yet known about the replication process in the mt DNA of parasitic nematodes , the high A+T content and the predicted structure of the AT-rich NCRs suggest an involvement in the initiation of replication 41 ., A substantial level of nucleotide difference ( 32 . 9% ) was detected in the complete mt genome between an individual of human-Trichuris and pig-Trichuris from China ., The sequence variation detected in the 13 protein-coding genes ( 25 . 4–47 . 4% ) and in NCRs ( 36 . 6% ) was consistent with previous findings of variation in the nucleotide sequences of the nuclear ITS rDNA from human and pig 14 ., However , for many nematodes 42 , 43 , there is usually greater within-species variation in mt protein-coding genes than in the ITS ., For example , the magnitude of the nucleotide sequence variation in the 12 common mt protein genes ( 3–7% ) 20 was greater than the 15 ( 1 . 8% ) variable positions in the ITS ( over 852 bp ) detected among multiple individuals of the human hookworm , N . americanus 44 ., Comparison between human- and pig-derived Trichuris from China also revealed variation at 1299 amino acid positions in the 13 predicted mt protein sequences ., This level of amino acid variation ( 36 . 4% ) is very high , given that mt proteins are usually conserved within a species due to structural and functional constraints 45 ., In addition , previous studies of other nematodes have detected little to no within-species variation in protein sequences ., For example , no within-species variation was detected in a COX1 region of 131 amino acids for N . americanus and for related hookworms , including A . caninum Ancylostoma and A . duodenale 24 , 33 ., Similarly , amino acid substitutions were recorded at only two of 196 ( 1% ) positions ( based on a comparison of conceptually translated sequences originating from GenBank accession nos . AF303135-AF303159 ) in partial COX1 among 151 N . americanus samples from four locations in China 46 ., In the present study , the greatest numbers of amino acid differences between human-Trichuris and pig-Trichuris were in the NAD4 ( n\u200a=\u200a235; 40 . 9% ) , NAD5 ( n\u200a=\u200a220; 35 . 9% ) and NAD2 ( n\u200a=\u200a120; 34 . 1% ) sequences; these percentages were significantly higher than that ( 4 . 9–10% ) between the two hookworms A . caninum and A . duodenale 24 , 33 ., The nature , extent and significance of the amino acid sequence variation between Trichuris from the human and pig hosts and from different geographical origins needs to be evaluated further , because there is virtually no published data on the magnitude of within-species variation in mt protein sequences for members of the genus Trichuris ., Genetic variation between human- and pig-Trichuris was also detected here in the two mt ribosomal RNA gene subunits ( rrnL and rrnS ) ., These subunits are usually more conserved in sequence than the protein genes 45 , which is also supported by the present data ., Comparison of the complete mt genomic data set between the two Trichuris individuals ( Ttr2 and TsCS1 ) displayed less sequence variation in rrnS and rrnL ( 24 . 6% and 25 . 1% ) compared with most protein genes ( 25 . 4–47 . 4% ) and the non-coding regions ( 36 . 6% ) ( Table 5 ) ., A region ( ∼430 bp ) in the conserved rrnL gene was used to examine the magnitude of genetic variation in Trichuris between the two different host species ., A comparison of the partial rrnL sequences among 16 Trichuris individuals revealed 89 ( 20 . 7% ) variable positions between human-Trichuris and pig-Trichuris , which is comparable with previous findings of a significant genetic difference ( 17% ) in nuclear ITS between the two operational taxonomic units ( OTUs ) in Spain 14 ., Taken together , the molecular evidence presented here supports the hypothesis that the gene pools of human-Trichuris and pig-Trichuris have been isolated for a substantial period of time and that they represent distinct species ., In spite of the genetic distinctiveness recorded here between them , host affiliation is not strict 47 ., Cross-infection of Trichuris between humans and pigs ( both directions ) has been described , but infection in the heterologous host is usually abbreviated 47 ., In spite of the compelling evidence of genetic distinctiveness between Trichuris specimens from human and pig hosts , interpretation from this study needs to be somewhat guarded until detailed population genetic investigations have been conducted ., Future studies could, ( i ) explore , in detail , nucleotide variation in ribosomal and mt DNAs within and among Trichuris populations from humans and pigs from a range of different countries employing , for example , mutation scanning-coupled sequencing 27 ,, ( ii ) establish , using accurate molecular tools , whether there is a particular affiliation between Trichuris and host in endemic regions and whether cross-host species infection is common or not , and, ( iii ) attempt to establish an experimental infection of Trichuris of human origin in pigs , in order to be able to investigate the genetic and reproductive relationships between human-Trichuris and pig-Trichuris ., Moreover , given the advent of high throughput genomic sequencing technologies , and the recent success in sequencing the nuclear genomes of the parasitic nematodes , B . malayi 48 and Ascaris suum 49 , it is conceivable that the genomes of human-Trichuris and pig-Trichuris will be characterized in the near future ., The transcriptome , and inferred proteome , characterised recently 50 will assist in future efforts to decode these genomes ., Such work will pave the way for future fundamental molecular explorations and the design of new methods for the treatment and control of one of the worlds socio-economically important nematodes 3 ., This focus is important , given the impact of Trichuris and other soil-transmitted helminths ( STHs ) , which affect billions of people and animals world-wide ., Although Trichuris species are seriously neglected , genomics and related approaches provide new opportunities for the discovery of novel intervention strategies , with major implications for improving animal and human health and well being globally ., In addition , the implications of genomic studies could also be highly relevant in relation to finding new treatments for immune-pathological diseases of humans 50 ., Interestingly , various studies 51–55 have indicated that iatrogenic infections of human patients suffering from immunological disorders ( such as inflammatory bowel disease , IBD ) with nematodes , such as pig-Trichuris eggs can significantly suppress clinical symptoms ., Although the mechanisms by which Trichuris modulates the human immune system are still unclear 52 , 56 , 57 , studies have proposed that a modified CD4+ T helper 2 ( Th2 ) -immune response and the production of anti-inflammatory cytokines , such the interleukins ( IL- ) IL-4 and IL-10 , contribute to the inhibition of effector mechanisms 56 , 58 , 59 ., Therefore , detailed investigations of pig-Trichuris at the molecular level could provide enormous scope for studying immuno-molecular mechanisms that take place between the parasite and humans affected by autoimmune or other immune diseases ., The mt genetic markers defined in the present study should be useful to verify the specific identity of Trichuris employed in such studies . | Introduction, Materials and Methods, Results, Discussion | The whipworm , Trichuris trichiura , causes trichuriasis in ∼600 million people worldwide , mainly in developing countries ., Whipworms also infect other animal hosts , including pigs ( T . suis ) , dogs ( T . vulpis ) and non-human primates , and cause disease in these hosts , which is similar to trichuriasis of humans ., Although Trichuris species are considered to be host specific , there has been considerable controversy , over the years , as to whether T . trichiura and T . suis are the same or distinct species ., Here , we characterised the entire mitochondrial genomes of human-derived Trichuris and pig-derived Trichuris , compared them and then tested the hypothesis that the parasites from these two host species are genetically distinct in a phylogenetic analysis of the sequence data ., Taken together , the findings support the proposal that T . trichiura and T . suis are separate species , consistent with previous data for nuclear ribosomal DNA ., Using molecular analytical tools , employing genetic markers defined herein , future work should conduct large-scale studies to establish whether T . trichiura is found in pigs and T . suis in humans in endemic regions . | Trichuriasis is a neglected tropical disease ( NTD ) caused by parasitic nematodes of the genus Trichuris ( Nematoda ) , causing significant human and animal health problems as well as considerable socio-economic consequences world-wide ., Although Trichuris species are considered to be relatively host specific , there has been significant controversy as to whether Trichuris infecting humans ( recognized as T . trichiura ) is a distinct species from that found in pigs ( recognized as T . suis ) , or not ., In the present study , we sequenced , annotated and compared the complete mitochondrial genomes of Trichuris from these two hosts and undertook a phylogenetic analysis of the mitochondrial datasets ., This analysis showed clear genetic distinctiveness and strong statistical support for the hypothesis that T . trichiura and T . suis are separate species , consistent with previous studies using nuclear ribosomal DNA sequence data ., Future studies could explore , using mitochondrial genetic markers defined in the present study , cross-transmission of Trichuris between pigs and humans in endemic regions , and the population genetics of T . trichiura and T . suis . | veterinary diseases, biology, zoology, veterinary science | null |
journal.pgen.1005858 | 2,016 | Expression Quantitative Trait Locus Mapping Studies in Mid-secretory Phase Endometrial Cells Identifies HLA-F and TAP2 as Fecundability-Associated Genes | Natural variation in fertility traits is heritable in humans 1 , yet identifying genes contributing to these traits remains challenging ., Although genome-wide association studies ( GWAS ) have identified variants associated with many other complex phenotypes , its application to fertility traits is challenging ., In particular , widespread contraceptive use among fertile couples and significant clinical heterogeneity among infertile couples makes it difficult to sample large numbers of fertile subjects with unprotected intercourse or infertile subjects whose inability to conceive results from the same underlying biological processes ., Although a few GWAS of male fertility 2 or infertility 3 , 4 traits have identified promising candidate genes , to date there have been no such studies in women ., To address these limitations , we have focused our genetic studies of fertility on members of a founder population , the Hutterites 1 , 2 , 5–8 ., This communal living group of European ancestry has limited contraceptive use , a uniform desire for large families , a prohibition of smoking , and fertility rates that are among the highest ever reported 9 , 10 ., For example , only 2% of Hutterite couples are childless 9 compared to 10–15% of the general population 11 ., Whereas miscarriages of clinically recognized pregnancies among Hutterite couples is 15 . 6% 8 , nearly identical to estimates of clinically recognized miscarriage rates in outbred populations 12 , recurrent miscarriages in childless couples are rare ( 0 of 525 interviewed Hutterite women 1 ) compared to 5% in the general population 13 ., Moreover , their communal lifestyle ensures that sociocultural factors influencing fertility are relatively uniform among Hutterite couples 1 , 14 ., We have proposed that their naturally high fertility rates , their reduced variance in environmental and lifestyle factors , and their limited gene pool due to the founder effect make the Hutterites an ideal population in which to dissect the genetic architecture of reproductive traits 1 , 2 , 15 ., Here , we use an integrated strategy that first identifies a set of candidate regulatory single nucleotide polymorphisms ( SNPs ) in mid-secretory phase endometrium by expression quantitative trait locus ( eQTL ) mapping in non-Hutterite women with two or more previous miscarriages ., We then tested for associations between those putatively functional expression ( e ) SNPs and fecundability in Hutterite women who are participants in a prospective study of pregnancy outcome 7 , 8 , and replicated the significant findings in an independent sample of women 16 ., We report here the discovery of independent associations between SNPs that are eQTLs for the HLA-F and TAP2 genes in mid-secretory phase endometrium and fecundability ( the probability of achieving pregnancy ) , thereby implicating maternal HLA region genes for the first time in implantation processes ., We performed eQTL mapping in the mid-secretory phase endometrium , corresponding to the luteal phase of the ovarian cycle , from 53 women with two or more early pregnancy losses , using 378 , 362 common ( ≥10% ) SNPs that were within 200kb of one or more of the 10 , 191 genes detected as expressed in these tissues ( i . e . , cis-eQTLs ) ( see Methods ) ., We observed 423 cis-eQTLs ( 416 unique SNPs ) for 132 genes at a false discovery rate ( FDR ) of 1% ( S1 Dataset ) ., We next looked for gene ontology enrichments in the genes associated with eQTLs at a FDR of 1% using DAVID 17 , 18 and GREAT 19 ., We found an enrichment of the GO Biological Process of “antigen processing and presentation” ( DAVID , FDR 2 . 5x10-5 ) , the GO Molecular Function of “MHC class 1 receptor activity” ( DAVID , FDR 5 . 30x10-4 ) , and GO Cellular Component “MHC class 1 protein complex” ( GREAT , FDR 2 . 02x10-6 ) ., Many of the DAVID and GREAT enrichments overlapped , and both highlighted the importance of immune related genes among those with eQTLs in mid-secretory phase endometrium ., To assess the clinical relevance of these eQTLs on female fertility traits , we first pruned the 416 SNPs for strong LD ( r2 ≥ 0 . 95 in the Hutterites ) and then carried forward 245 expression ( e ) SNPs for association studies in the prospective study participants ., We previously genotyped 208 of the 327 Hutterite women in a prospective study of pregnancy outcomes 6 , 7 , 15 , 20 with the Affymetrix 500k or 6 . 0 genotyping chips ., Using these genotypes as framework markers , we imputed all variants present in the whole genome sequences of 98 Hutterites to these women using PRIMAL , an imputation program that utilizes both pedigree- and LD-based imputation to provide on average of 87% coverage and >99% accuracy in the Hutterite pedigree 21 ., From among the 245 ( LD-pruned ) SNPs that were eQTLs at a FDR 1% , genotypes for 189 ( associated with the expression of 108 genes ) were known for at least 85% of Hutterite women in the study of fecundability ., We compared the length of the intervals from the resumption of menses after a prior pregnancy or miscarriage or following the discontinuation of birth control use ( referred to as time0 ) to a positive pregnancy test in women who were not nursing at time0 of each included interval ., For first pregnancies , we considered the length of the interval from the first menses after marriage to a positive pregnancy test ( see Methods ) ., If pregnancy occurred prior to the resumption of menses ( or before the first period after marriage in first pregnancies ) , we considered the interval to be 14 days ., Data were available for 191 intervals in 117 women ( see Methods ) ; 178 of the observed intervals resulted in a pregnancy at the time of the last follow-up ., We used life-table analysis to compare intervals between genotype classes , adjusting for two significant covariates: maternal age and number of prior births ( classified as 0–1 , 2–3 , or ≥4 ) ( see Methods ) ., Among the 189 eSNPs that we examined , genotypes for 21 were associated with the length of the interval to pregnancy at a P < 0 . 05 ( Table 1 ) ., Two of these eSNPs were significant at a FDR of 5% and after Bonferroni correction for 189 tests ., The most significant association was with rs2071473 , an eSNP associated with expression of the TAP2 gene in the HLA class II region; the C allele at this SNP was associated with longer intervals to pregnancy and higher expression of TAP2 gene in mid-secretory phase endometrium ( Fig 1 ) ., The median interval lengths to pregnancy were 2 . 0 ( lower , upper quartile 1 . 2 , 4 . 7 ) , 3 . 1 ( 1 . 9 , 6 . 2 ) , and 4 . 0 ( 2 . 0 , 7 . 6 ) months among women with the TT , CT , and CC genotypes , respectively , at this eSNP ( P = 1 . 3x10-4 ) ., The second association was with rs2523393 , an eSNP associated with expression of the HLA-F gene in the HLA class I region; the G allele at this SNP was associated with longer intervals to pregnancy and lower expression of HLA-F in mid-secretory phase endometrium ( Fig 2 ) ., The median interval lengths to pregnancy were 2 . 3 ( 1 . 8 , 4 . 5 ) , 2 . 6 ( 1 . 4 , 4 . 8 ) , and 4 . 9 ( 2 . 0 , 11 . 7 ) months among women with the AA , AG , and GG genotypes , respectively , at this eSNP ( P = 4 . 0x10-4 ) ., For both eSNPs , intervals were longer and genotype differences more pronounced among women at lower parity ( S1 and S2 Figs ) ., Among the 21 eSNPs with P <0 . 05 , nine ( 43% ) were associated with expression of HLA region genes: one with TAP1 , three with HLA-F , three with HLA-G , and two with MICA , consistent with the gene ontology analysis identifying enrichments for genes with antigen processing and presentation functions among those with eQTLs in mid-secretory phase endometrium ., The results for all 189 eSNPs and their associated genes are shown in S2 Dataset ., Previous studies of HLA and fertility in the Hutterites have shown that HLA matching between partners for alleles at the class II locus HLA-DRB1 is associated with reduced fecundability , presumably due to the higher risk for class II compatible embryos among these couples 7 ., To rule out that maternal-fetal compatibility at the TAP2 or HLA-F locus accounts for the observed effects in this study we repeated the fecundability analysis , first stratifying couples based on husband’s genotype ( rather than wife’s genotype ) and then stratifying couples based on the wife’s genotype ( as above ) but now including the husband’s genotype as a covariate ., We reasoned that if longer intervals are due to maternal-fetal compatibility and not maternal genotypes per se , then results of analyses stratifying by husband’s genotype should yield results similar to analyses stratified by wife’s genotype , and analyses including both husband’s and wife’s genotypes should be more significant than analyses considering either one individually ., Neither eSNP was significant in the analysis considering the husband’s genotype as a main effect on length of intervals to pregnancy ( HLA-F rs2523393 P = 0 . 94; TAP2 rs2071473 P = 0 . 56 ) ., When husband’s genotype was included as a covariate in the model , the P-values were reduced from 4 . 0x10-4 to 0 . 0014 for rs2523393 and from 1 . 3x10-4 to 0 . 0015 for rs2071473 , but the effect size associated with the risk alleles remained largely unchanged ( β coefficients changed from 0 . 39 to 0 . 34 for rs2523393 and -0 . 44 to -0 . 36 rs2071473 when husbands’ genotypes were included as a covariate ) ., These data indicate that maternal genotype at these two eSNPs is driving the association with time to pregnancy; there is no evidence for paternal or fetal genotype effects at these eSNPs contributing to interval lengths ., Although these two eSNPs reside at opposite ends of the HLA region and are separated by ~3Mb , there are moderate levels of LD between them in the Hutterites ( r2 = 0 . 19 ) ., To determine the statistical independence of the associations with fecundability , we repeated the time to pregnancy analysis but included the genotype at the other eSNP as a covariate ., In the analysis of rs2071473 ( TAP2 ) that included genotype at rs2523393 ( HLA-F ) as a covariate , the effect size and P-value changed from β = 0 . 39 ( P = 1 . 3x10-4 ) to β = 0 . 23 ( P = 0 . 0064 ) ; in the analysis of the rs2523393 ( HLA-F ) that included genotype at rs2071473 ( TAP2 ) as a covariate , the effect size and P-value changed from β = -0 . 44 ( P = 4 . 0x10-4 ) to -0 . 34 ( P = 0 . 047 ) ., Thus , while the magnitude of each association is reduced in the analyses conditioning on the alternate eSNP , both retain independent effects on fecundability ., The observed reduction in β values and significance is likely due to the LD between the SNPs ., To further examine this , we stratified the women into three groups based on being homozygous at both , one , or neither of the high risk ( longer interval ) alleles at each eSNP ( CC at rs2071473 TAP2 and GG at rs2523393 HLA-F ) ( Fig 3 ) ., If the effects at these two loci were independent , then women who are homozygous for the high risk allele at both eSNPs should have longer intervals than women who are homozygous at only one or neither high risk allele ., Indeed , intervals to pregnancy were longest among women homozygous for both rs2071473-CC and rs2523393-GG ( median interval 5 . 2 months 1 . 9 , 11 . 0 months ) , intermediate among homozygous for only one of the high risk alleles ( median interval 4 . 0 months 2 . 1 , 9 . 3 months , and shortest among women who were not homozygous for either high risk allele ( median interval 2 . 4 months 1 . 4 , 4 . 8 months ) ( P = 2 . 9x10-4 ) ., Moreover , women homozygous for the risk alleles at both the TAP2 and HLA-F eSNPs had significantly longer intervals compared to women who were homozygous for a risk allele at only one of the two eSNPs ( P = 1 . 6x10-5 ) ., Taken together these analyses indicate that the TAP2 and HLA-F associations are independent and have additive effects on fecundability ., Using the same approach as that used in the Hutterites , we first examined the genotype effects of each SNP on fecundability and then the joint effects of the combined genotypes at each locus ., At rs2071473 , the TAP2 eSNP , the median interval lengths to pregnancy were 5 . 0 ( 4 . 0 , 7 . 0 ) , 6 . 0 ( 4 . 0 , 8 . 2 ) , and 6 . 0 ( 4 . 0 , 9 . 0 ) months among women with the TT , CT , and CC genotypes , respectively ( P = 0 . 083; Fig 4A ) ., At rs2523393 , the HLA-F eSNP , the median interval lengths were 5 . 0 ( 4 . 0 , 8 . 5 ) , 6 . 0 ( 4 . 0 , 8 . 0 ) , and 6 . 0 ( 5 . 0 , 10 . 0 ) months among women with the AA , AG , and GG genotypes , respectively ( P = 0 . 155; Fig 4B ) ., Although these results did reach nominal significance in the RFTS cohort , the 95% confidence intervals of the ORs in the Hutterites and RFTS cohort overlap ( S1 Table ) ., In the combined analysis , intervals to pregnancy were longest among women homozygous for both rs2071473-CC and rs2523393-GG ( median interval 8 . 0 months 5 . 5 , 12 . 5 months ) , intermediate among homozygous for only one of the high risk alleles ( median interval 6 . 0 months 4 . 0 , 9 . 0 months , and shortest among women who were not homozygous for either high risk allele ( median interval 5 . 0 months 4 . 0 , 7 . 0 months ) ( P = 0 . 033; Fig 4C ) , as we observed in the Hutterites ., Because there are many SNPs in strong LD with our lead eSNPs in the Hutterites , it cannot be inferred from association studies which of these linked SNPs are the true causal variants ., To address this question , we used in silico analyses to determine which of the fecundability-associated eQTLs are in or near regulatory elements in decidualized human endometrial stromal cells 22 , 23 , ENCODE-annotated functional sites in the endometrial cell lines ECC-1 and Ishikawa 24 , as well as the complete ENCODE regulatory element dataset ., To interrogate more completely the variation in these regions , we used whole genome sequence data in the Hutterites to survey all variation in the 500kb windows flanking each of the two eSNPs that were associated with fecundability ., After filtering our variants with minor allele frequencies <0 . 10 and call rates <85% , 4 , 442 variants remained in the TAP2 region and 2 , 675 variants remained in the HLA-F region ., We then filtered these variants based on their LD with each lead eSNP and retained the 70 variants with LD r2 ≥ 0 . 7 with rs2071473 and the 62 variants with LD r2 ≥ 0 . 7 with rs2523393 ., The variants that had LD r2 ≥ 0 . 7 with the lead eSNP in each region defined an approximately 6kb window around the lead SNP ., We repeated the association studies with all variants within each 6kb window and fecundability ., Because many of these variants were not included in the eQTL study in the 53 women with recurrent early pregnancy loss , we also imputed the missing genotypes using whole genome sequences from 100 European American individuals 25 ( see Methods ) and performed eQTL mapping in the mid-secretory phase endometrial RNA using these variants , as described above ., The lead eSNP at the TAP2 locus , rs2071473 , is located within an intron of the HLA-DOB gene and is near ( ~600bp ) an NF2R2 transcription factor ( TF ) binding site in hESC; EP300 , FOXM1 , ATF2 , and RUNX3 binding sites in a B-lymphoblastoid cell line ( GM12878 ) ; and a DNase-I hypersensitivity site in 40 ENCODE cell lines ( Fig 5A ) ., This SNP is also within ~800bp of a FAIRE peak in hESC ., Another SNP , rs2856995 , that is 732bp upstream of and in perfect LD ( r2 = 1 ) with rs2071473 resides within the NF2R2 binding site and 92bp upstream of the FAIRE site in hESC ., Multiple other variants are located adjacent to FAIRE sites and among the 36 of these variants with eQTLs , 34 were eQTLs only for TAP2 ( FDR <15% ) , and two were eQTLs for TAP1 ( FDR = 14% ) ( S3 Dataset ) ., The lead eSNP at the HLA-F locus , rs2523393 , is located within an intron of HLA-F-AS1 , an antisense transcript that was not expressed in mid-secretory endometrium ., The eSNP is about 10kb downstream of HLA-F , and in perfect LD ( r2 = 1 ) with a cluster of SNPs ~360bp away that reside within multiple ENCODE-annotated functional sites in endometrial cell lines 26 ( Fig 5B ) ., One such SNP , rs2523389 , is in a DNaseI hypersensitivity site in 120 ENCODE cell lines including the endometrial derived cell lines ECC-1 and Ishikawa treated with 10nM estradiol ., This variant is also in a CTCF binding site that is present in 97 ENCODE cell lines , including the endometrial cell line ECC-1 , and in a c-Myc binding site in a leukemia cell line ., Another variant , rs2523391 , is in a STAT3 binding site in a mammary gland cell line and 6bp from a CEBPB TF ChIP-seq binding site in HeLa and HepG2 cell lines , in addition to the same functional sites as rs2523389 ., Although the c-Myc , STAT3 , and CEBPB binding sites were not present in the hESC or ENCODE endometrial cell lines , these transcription factors are essential for decidualization of endometrial stromal cells and the successful establishment of pregnancy 27–29 ., Among the variants in LD with rs2523393 at r2 ≥ 0 . 7 and with eQTL results , all were most strongly associated with expression of HLA-F ( S4 Dataset ) ., One eSNP in the HLA-F promoter ( rs1362126; r2 = 0 . 78 to rs2523393 ) was also an eQTL for HLA-G ( FDR <1% ) , although to a lesser degree than for HLA-F ( HLA-F eQTL P = 2 . 18x10-6 , HLA-G eQTL P = 2 . 06x10-3 ) ., Overall , these data indicate that our lead eSNPs and/or a small number of variants in perfect LD with those eSNPs are plausible causal candidates for the observed associations with fecundability in each region ., The mechanisms that allow the fetal allograft to avoid maternal immunologic rejection and survive over relatively long gestational periods in placental mammals are still incompletely known , although our understanding of these processes have advanced considerably since Medawar proposed this paradox over 60 years ago 30 ., In particular , it has become clear that major histocompatibility complex ( MHC ) antigens , which play a central role in the rejection of non-self tissues , also contribute to maternal tolerance of the fetus , which is maintained in normal pregnancies ., For example , our group previously demonstrated that matching of HLA antigens ( the human MHC loci ) between Hutterite couples is associated with longer intervals from marriage to each birth compared to couples not matching for HLA 31 , and that longer intervals resulted from both higher miscarriage rates among couples matching for class I HLA-B antigens 8 and longer intervals to pregnancy among couples matching for class II HLA-DR antigens 7 ., More recently , we reported associations between maternal HLA-G genotypes and miscarriage 6 ., We and others have also shown associations between maternal or fetal HLA-G genotypes with recurrent pregnancy loss and preeclampsia 32–42 ., Finally , recent studies have elegantly demonstrated that two HLA that are expressed by fetal extravillous cytotrophoblast ( EV-CTB ) cells at the maternal-fetal interface , HLA-G and HLA-C , are ligands for inhibitory receptors ILT2/IL4 and KIR2DL3 , respectively , on immune cells 43–45 ., Collectively these data indicate that multiple HLA molecules play important independent roles at the maternal-fetal interface in human pregnancy and that their effects can influence pregnancy outcome throughout gestation ., In this study , we hypothesized that perturbations of genes expressed in mid-secretory phase endometrial cells could affect implantation and be visible as delayed time to pregnancy in otherwise fertile couples ., Although our study was unbiased with respect to genome location because we interrogated variation that was first identified through a genome-wide eQTL study , the results of both the eQTL study and the subsequent study of fecundability highlighted the importance of HLA region genes in achieving pregnancy ., Among the eQTLs taken forward to studies of fecundability in the Hutterites , nine ( 43% ) of the eSNPs with association P-values <0 . 05 were eQTLs for HLA region genes compared to 15% of all 189 eSNPs tested ., Two of the associations with fecundability were significant at a FDR of 5% , remained significant after correction for multiple testing , and were replicated in an independent sample of fertile women: one SNP was with an eQTL for TAP2 and one for HLA-F ., To our knowledge neither of these two HLA region genes has previously been directly implicated in pregnancy processes ., We further demonstrated that eQTLs for these two HLA loci in mid-secretory phase endometrium are independently associated with fecundability in fertile women and that neither paternal nor fetal genotype at these loci contributed to these effects ., These findings may be particularly relevant to women with primary infertility of unknown etiology , with recurrent implantation failure following in vitro fertilization ( IVF ) , or possibly even with recurrent early pregnancy loss ., Both genes are intriguing candidates for fecundability genes ., The TAP2 gene in the class II HLA region encodes the antigen peptide transporter 2 protein ., TAP2 forms a heterodimer with TAP1 ( encoded by the TAP1 gene , located 7kb away ) in order to transport peptides from the cytoplasm to the endoplasmic reticulum , where they are loaded into assembling class I HLA molecules prior to their transport to the cell surface ., The association of TAP complex with HLA class I molecules , including HLA-F 46 , is critical for their expression on the cell surface 47 ., HLA-F , which is located ~3Mb telomeric to TAP2 , encodes a class I HLA protein that is considered “non-classical” because it has limited coding polymorphisms and restricted tissue distribution 48 , and functions that are still poorly characterized but likely distinct from the classical class I HLA ( HLA-A , HLA-B , HLA-C ) ., In fact , recent studies have shown that HLA-F physically interacts with the KIR3DL2 and KIR2DS4 receptors on natural killer ( NK ) cells 49 , an abundant and critical cell in the maternal uterus that proliferates during the secretory phase and then throughout pregnancy 50 ., Later in pregnancy , HLA-F is expressed in EV-CTB 51–54 , although its function in placental cells is not well characterized 52 ., Our combined results suggest that perturbations in expression of either gene in endometrial cells in the mid-secretory phase influences implantation success , with overexpression of TAP2 and underexpression of HLA-F resulting in delayed time to pregnancy ., We found multiple variants in perfect LD with both eSNPs that reside in transcription factor binding sites and other regulatory elements in endometrial cell lines ., The TAP2-associated variants are located within a NR2F2 ( COUP-TFII ) binding site ., Multiple studies have shown that female mice deficient in NR2F2 have implantation failure , with impairments of both embryo attachment and uterine decidualization 55–57 , and NR2F2 knock downs a human endometrial stromal cell line significantly reduces TAP2 expression 22 ., These combined data suggest a potential mechanism for the association we observed with expression of TAP2 and fecundability in women ., At the HLA-F locus , variants associated with gene expression level and fecundability are in DNaseI hypersensitivity sites , a marker of open chromatin and transcriptional activity , in multiple human endometrial cell lines 26 , suggesting that one or more of these variants may indeed be causally associated with both gene expression and fecundability ., Although genome-wide association studies can be powerful approaches for identifying susceptibility loci for common diseases and complex phenotypes , they require very large sample sizes that may be infeasible to acquire for many important phenotypes ., We used an alternative approach for mapping fecundability genes by first reducing the search space to SNPs that were associated with gene expression in a relevant tissue and then taking this smaller set of regulatory SNPs forward to an association study in carefully phenotyped subjects ., This approach revealed two novel associations with fecundability and immediate intuition regarding the genes underlying each association , the relationship between gene expression and fecundability , and potential mechanisms for these associations ., Future studies will be required to characterize the role of these molecules in the implantation process and to evaluate their potential as drug targets for treatment of conditions related to suboptimal implantation ., Fifty-eight women underwent endometrial biopsies as part of their clinical evaluation for recurrent pregnancy loss at the University of Chicago , after obtaining informed consent ., These women were between the ages of 26 and 43 years and had at least two previous pregnancy losses before10 weeks gestation ., Fifty-two ( 90% ) were of European ancestry , two ( 3% ) were of Asian ancestry , and four ( 7% ) were of African ancestry ., Medical records and individual diagnoses were not available to us for this study , and all women were included in the expression studies ., Because recurrent miscarriage can result from many potential causes , and nearly half remain unexplained after evaluation , we reasoned that gene expression in the endometrium from these women would maximize variation in gene expression and increase our power to detect eQTLs ., We were unable to obtain samples from women without a history of pregnancy loss for this study ., Endometrial biopsies were performed during the mid-secretory phase ( 9–11 days after endogenous luteinizing hormone LH surge , detected by each woman testing her daily urine ) and immediately frozen on dry ice; samples were stored at -80°C until RNA was extracted , as previously described 58 ., Histological examination of the biopsies confirmed endometrial tissue from the fundus of the uterus; endometrial glands and epithelium were present ., RNA was extracted from the endometrial biopsies using a phenol-chloroform phase separation with TRIzol per the manufacturer’s directions ( Life Technologies Corp . , Carlsbad , CA , USA ) and RNeasy RNA extraction kit ( Qiagen , Venlo , Netherlands; per manufacturer’s directions ) ., RNA quality was assessed using the Agilent 2100 Bioanalyzer ( Agilent Technologies , Santa Clara , CA ) ., The average RNA integrity number ( RIN ) score 59 was 8 . 06 ( range 6 . 7–9 . 3 ) ., We were unable to obtain RIN scores for two samples due to extremely low concentration ., However , these two samples passed all gene expression QC and were therefore included in the eQTL mapping studies ., Gene expression was measured in RNA from 58 individuals ., We included triplicate samples for three women and duplicate samples for 29 ., Gene expression was interrogated using the Human HT12v4 Expression BeadChip ( Illumina , Inc . , San Diego , CA , USA ) , which contains 47 , 231 probes that target 11 , 121 unique RefSeq genes ., cDNA synthesis , hybridization , scanning and image processing and returned probe intensity measurements were performed at the University of Chicago Functional Genomics Core ., Intensity estimates were log-transformed and quantile normalized using the ‘lumi’ package in R 60 ( see S3 Fig ) ., To remove probes for targets that were likely not expressed , all probes that did not have a detection P-value <0 . 05 in at least 70% of the samples were removed ( leaving 17 , 208 probes ) ., We further removed probes that did not map uniquely to the HG19 genome using Burrows-Wheeler Aligner ( BWA ) , and probes that contained CEU HapMap SNPs with the QC+ designation ( 4 , 825 probes total excluded ) ., After quality control ( QC ) , 12 , 383 probes remained and were included in our eQTL studies ., Probe averages were taken for replicate samples ., Of the 11 , 121 unique genes targeted on the array , 2 , 192 ( 20% ) had multiple probes 61 ., For those genes with multiple probes , we chose the most 3’ probe in the gene to estimate expression ., Samples from two women were excluded prior to QC because there were too few probes detected after hybridization ( 13 , 189 and 15 , 863 probes , respectively , compared to a median count of 21 , 341 out of 47 , 231 probes on the array ) ; and three women were excluded prior to analysis because there was no DNA available for one and low genotyping call rates in two ( see below ) ., The remaining 53 women ( 49 European ancestry , 2 Asian ancestry , 2 African American ancestry ) had both high quality expression and genotype data ., In those 53 samples , 10 , 191 genes were detected as expressed ., Processing batch and array were significantly associated with the variance in gene expression based on principle component ( PC ) analysis and their effects were regressed out using a linear model ( see S4 Fig ) ., Other covariates that were considered ( age , BMI , race , and season of biopsy collection ) were not significant in these samples ( see S2 Table ) ., An overview of the sample inclusion pipeline is shown in S5A Fig . DNA from women in the gene expression studies were genotyped with the Affymetrix Axiom Genome-Wide CEU 1 Array at the UCSF Genomics Core Facility ., We performed QC checks using PLINK 62 , and removed 4 , 922 SNPs with <95% genotype call rates , 503 with Hardy-Weinberg P-values ≤0 . 001 , 336 non-autosomal SNPs , and 252 , 872 with minor allele frequencies <0 . 10 ., There were 370 , 008 SNPs remaining ., After excluding two women with low call rates , the remaining 53 subjects had SNP call rates >97% ., Linear regression was used to test for associations between the expression levels of 10 , 191 genes and genotypes at the 378 , 362 SNPs with a minor allele frequency greater than 10% , using the R package Matrix eQTL 63 ., Genotypes were recoded as 0 , 1 , or 2 to reflect an additive model ., To maximize the power in our sample , we tested for associations only with SNPs within ~200kb of the transcription start site of a gene , a distance that would include nearly all cis regulatory SNPs 64–66 ., The Matrix eQTL package assigns both a p-value and a false discovery rate ( FDR ) to each SNP-gene association ., The FDR was calculated using the Benjamini and Hochberg 67 procedure ( see 63 for detailed methods ) ., eQTL mapping was additionally performed including only the 49 women of European ancestry ., As expected , p-values were generally less significant in the smaller sample , however the HLA-F and TAP2 eQTLs remained significant at a FDR <1% ( S5 Dataset ) ., We used the Database for Annotation , Visualization and Integrated Discovery ( DAVID ) v6 . 7 17 , 68 to interrogate pathway and gene enrichment for eQTLs at a FDR 1% compared to all gene-SNP combinations in our analysis ( background ) ., An enrichment score was calculated using Fishers Exact test ( modified as EASE score ) on gene count for eQTLs at a FDR of 1% compared to all genes tested ., We used the high classification stringency for our analysis ., We also used the Genomic Regions Enrichment of Annotations Tool ( GREAT ) to analyze the significance of SNPs which are eQTLs at a FDR of 1% 19 ., GREAT first associates genomic regions with nearby genes and then applies the functional annotations for those genes to the regions ., We used the basal plus extension definition of a gene regulatory domain , in which a gene’s defined regulatory domain expands until it reaches the nearest gene’s basal domain or maximally 5kb upstream and 1kb downstream ., Using this definition , SNPs located between two genes may include both gene regulatory domains ., GREAT uses a hypergeometric test over these defined genomic domains to assess enrichment between foreground ( FDR 1% eQTLs ) and background ( all SNPs for which there is a gene-SNP pair tested in the eQTL analysis ) ., The data included in our analyses of fecundability were derived from a prospective study of pregnancy outcome in South Dakota Hutterites that was initiated in 1986 , as previously described 6–8 , 69 ., The women in this study are provided with calendar diaries and EPT pregnancy test kits ( Warner-Lambert Co . | Introduction, Results, Discussion, Materials and Methods | Fertility traits in humans are heritable , however , little is known about the genes that influence reproductive outcomes or the genetic variants that contribute to differences in these traits between individuals , particularly women ., To address this gap in knowledge , we performed an unbiased genome-wide expression quantitative trait locus ( eQTL ) mapping study to identify common regulatory ( expression ) single nucleotide polymorphisms ( eSNPs ) in mid-secretory endometrium ., We identified 423 cis-eQTLs for 132 genes that were significant at a false discovery rate ( FDR ) of 1% ., After pruning for strong LD ( r2 >0 . 95 ) , we tested for associations between eSNPs and fecundability ( the ability to get pregnant ) , measured as the length of the interval to pregnancy , in 117 women ., Two eSNPs were associated with fecundability at a FDR of 5%; both were in the HLA region and were eQTLs for the TAP2 gene ( P = 1 . 3x10-4 ) and the HLA-F gene ( P = 4 . 0x10-4 ) , respectively ., The effects of these SNPs on fecundability were replicated in an independent sample ., The two eSNPs reside within or near regulatory elements in decidualized human endometrial stromal cells ., Our study integrating eQTL mapping in a primary tissue with association studies of a related phenotype revealed novel genes and associated alleles with independent effects on fecundability , and identified a central role for two HLA region genes in human implantation success . | Little is known about the genetics of female fertility ., In this study , we addressed this gap in knowledge by first searching for genetic variants that regulate gene expression in uterine endometrial cells , and then testing those functional variants for associations with the length of time to pregnancy in fertile women ., Two functional genetic variants were associated with time to pregnancy in women after correcting for multiple testing ., Those variants were each associated with the expression of genes in the HLA region , HLA-F and TAP2 , which are have not previously been implicated female fertility ., The association between HLA-F and TAP2 genotypes on the length of time to pregnancy was replicated in an independent cohort of women ., Because HLA-F and TAP2 are involved in immune processes , these results suggest their role in specific immune regulation in the endometrium during implantation ., Future studies will characterize these molecules in the implantation process and their potential as drug targets for treatment of conditions related to implantation failure . | cell binding, cell physiology, uterus, medicine and health sciences, reproductive system, maternal health, obstetrics and gynecology, variant genotypes, alleles, genetic mapping, womens health, pregnancy, molecular biology techniques, pregnancy complications, research and analysis methods, endometrium, gene mapping, gene expression, miscarriage, molecular biology, genetic loci, anatomy, cell biology, heredity, genetics, biology and life sciences | null |
journal.ppat.1001224 | 2,010 | Identifying the Age Cohort Responsible for Transmission in a Natural Outbreak of Bordetella bronchiseptica | Containing and ultimately eliminating infectious disease remains a central goal for many animal and public health officials ., Dissecting disease transmission – in terms of identifying the routes and potentially heterogeneous rates of disease spread 1 – is an essential step in devising or optimizing intervention strategies aimed at pathogen eradication 1 , 2 ., This is because heterogeneities in transmission that arise due to for example age- or sex-specific differences among individuals 2 , 3 can greatly affect invasion and eradication criteria 1 ., Unfortunately , precise measurements of transmission remain elusive due to the immense difficulties associated with identifying the nature of a potential contact , the probability of infection given a contact 4 and important drivers of heterogeneities in transmission 2 , 3 , 5 ., A key reason for these difficulties is that transmission events are rarely observed directly , with some notable exceptions 6 ., One useful approach that can shed partial light on the transmission process is to measure the force-of-infection ( FOI: λ ) , or the per capita conversion rate of susceptible hosts 7 ., The simplest way to think about the FOI , is that over a short interval of time – say from time t to t+Δ – the probability that a disease negative individual becomes disease positive is λΔ ., The most popular way to estimate λ is through use of the observed age-specific prevalence ( or the proportion of individuals that are disease positive in a cross-sectional sample ) , due to the ease with which it is measured in most populations 8 ., Indeed , FOI estimates have been calculated from age-prevalence data for human , and to a lesser extent , wildlife infections 3 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 ., Estimating the FOI can help identify those age-classes responsible for transmission and evaluate the relative effects of each group on overall transmission ., Here we evaluate the relative effects of sex , age and social structure on the transmission dynamics of the respiratory pathogen Bordetella bronchiseptica within a commercial rabbitry of New Zealand White ( NZW ) rabbits ., In doing so we illustrate how analysis of age-prevalence data can be used to estimate the age-specific FOI ., The importance of social organization in B . bronchiseptica transmission is also considered ., To test for litter-based transmission events – for example , from mother to offspring or between siblings – we checked for significant correlation among the fate of siblings by using a litter-based random-effects binomial regression to estimate the importance of horizontal versus pseudovertical transmission 20 , 21 ., The statistical tools we employ here are general and can be applied to a range of directly transmitted medical and veterinary diseases to help shed light on the dynamics of disease spread and allow an assessment to be made of the best methods for effective long-term disease control ., The Bordetella genus contains three closely related gram-negative bacteria that cause respiratory infections in humans and other mammals 22 ., Whereas B . pertussis and B . parapertussis largely infect humans and cause the acute respiratory disease known as whooping cough 23 , B . bronchiseptica typically causes chronic infections in a wide range of mammals 24 ., Indeed B . bronchiseptica infection is often endemic in agricultural settings – including commercial rabbitries 25 , 26 – where rapid spread and persistent infection make it difficult to control 23 ., Despite its widespread nature , there is a paucity of data describing the epidemiology of B . bronchiseptica in terms of both the main route ( s ) of , and likely cohort ( s ) responsible for disease transmission ., As a respiratory infection , the major physical route of transmission is oral-nasal via direct aerosol droplets 27 , 28 ., Based on the published literature 26 , 27 , 29 , we propose a set of plausible routes of transmission within a commercial rabbitry would include: These ( not mutually exclusive ) possible routes of transmission may result in the prevalence of infection changing with age in different ways which can be related to different underlying hazard models/FOI patterns ( Figure 1 ) ., As we will explore below , each of these transmission possibilities translates into a specific prediction which can be tested using our statistical framework ., To identify the parsimonious hypothesis , we applied a piece-wise constant model for the age-specific FOI 9 , 11 , 15 ., Since B . bronchiseptica is an endemic persistent infection 23 , we used a catalytic framework ( which assumes a one-way flow from susceptible to infected ) ., The importance of sex and location ( facility building ) and time of sampling on FOI estimates was also determined ., We considered the importance of social mixing and organization in B . bronchiseptica transmission using random effect logistic regression estimates to control for litter as confounding variable in transmission models ., In parallel , we took a molecular epidemiological approach to investigate whether strain-specific differences existed in the epidemiological pattern of infection 30 , 31 ., All protocols involving rabbits were approved by the Institutional Animal Care and Use Committee ( IACUC ) at the Pennsylvania State University according to the guidelines of the American Association for Laboratory Animal Science ., This study was conducted at a commercial rabbitry which breeds NZW rabbits ., The rabbitry comprised of three separate animal breeding buildings ( Table 1; buildings A - C ) ., Background health checks – in the form of comprehensive monthly pathology reports testing for >17 pathogens – have been carried out since January 2003 ( n\u200a=\u200a2 to 4 rabbits/month/building ) ., These reports show that B . bronchiseptica has been endemic in the rabbitry since testing began and that of the other pathogens screened , only non-pathogenic Eimeria species ( intestinal coccidia ) are occasionally isolated ., Importantly , our rabbitry is Pasteurella multocida free – infection with this respiratory pathogen has long been associated with upper respiratory disease ( URD ) in rabbits 32 – with no URD reported in the last 30 years ., Kits were weaned at 4–5 weeks of age and ‘weanlings’ segregated by sex and co-housed in sibling pairs ., Rabbits of good breeding stock were selected as ‘future breeders’ and housed in pairs ., The remaining ‘stock’ rabbits were housed singly and sold at 8–10 weeks old ., ‘Breeder’ rabbits were initially bred at 5 or 6 months of age for females and males respectively ., Breeders were , housed individually and rebred when litters were weaned ., For all rabbits included in this study , the date of sampling , building and rabbit identification number were ascertained along with nasal swab ( BD sterile swab , product # 220518 ) ., A total of eight sampling efforts were carried out from November 2006 to September 2008 , culminating in the collective nasal swabbing of 602 rabbits total ., All nasal swabs were streaked onto Bordet-Gengou ( BG ) agar ( Difco ) containing 10% sheeps blood ( Hema Resources ) with 20 µg/mL streptomycin ( Sigma ) as soon as possible after collection and incubated at 35°C for ∼3–5 days ., The sampling strategies were as follows: We used Multi-Locus Sequence Typing ( MLST ) analysis 33 , 34 to determine the phylogenetic relationships among 90 B . bronchiseptica isolates from 4 sampling efforts across three rabbitry buildings , as previously described for Bordetella ., Briefly , genomic DNA from each isolate was obtained using a DNAeasy Tissue Kit ( Qiagen ) and nucleotide sequences were determined for internal regions of seven housekeeping genes for all 90 isolates ( see the Bordetella MLST database at http://pubmlst . org/bordetella ) ., All alleles were double stranded sequenced at The Pennsylvania State University Genomic Sequencing Center and an allele number was assigned to each unique allele sequence ., The combination of the allele numbers at the seven loci defines the sequence types ( ST ) or allelic profile of each strain 33 , 35 ., Two different catalytic models were fitted to the data: a piecewise constant FOI with 3 age-intervals ( Figure 2b ) , corresponding to the hypothesized routes of transmission outlined in the Introduction and Figure 1 and a constant FOI corresponding to the null hypothesis ., The 3 age-interval model fits the data better ( having the lowest AIC value of 219 . 39 ) than the null hypothesis of a constant FOI ( ΔAIC of model ( c ) versus ( a ) =\u200a39 . 6 ) ., The model results show close correspondence between the observed and expected prevalence data ( Figure 2a ) ., Both data and model fits exhibit a rapid increase in prevalence during the first and second age-classes ( i . e . in rabbits up to 5 months; Figure 2a ) ., During the first month of life , the estimated FOI is substantial ( Figure 2b; FOI\u200a=\u200a0 . 16 month−1 ) and peaks in the second age-class ( Figure 2b; FOI\u200a=\u200a0 . 20 month−1 ) , with the older age classes ( from 5–30 months of age ) having the lowest FOI estimates ( Figure 2b; from 3 . 3×10−4 to virtually zero ) ., Next we examined whether gender differences existed for FOI estimates ., No differences between sexes were found in the FOI estimates ., In the younger age-classes , the prevalence data and model estimates peaked in the second age-class resulting in positive FOI estimates in young weaned kits ( 1 to 4 month olds: Figure 3a–d ) ., In the older age-classes , B . bronchiseptica prevalence asymptoted and subsequently fell for both sexes , with a concomitant decline in the FOI estimates ( Figure 3a–d ) ., To examine the likelihood of becoming infected from an infected sibling , we ran a binomial regression on the experimental sibling-to-sibling transmission experiment ( see Sampling Strategy Two in M&M for details ) ., Being co-housed with an infected sibling increased the probability of becoming B . bronchiseptica positive ( Figure 4a: co-housed with infected sibling: Z\u200a=\u200a2 . 42 , p\u200a=\u200a0 . 016 ) , such that uninfected kits were 3 . 85 times more likely to become infected when they were co-housed with an infected- compared to an uninfected-kit ( Figure 4a: 95% C . I . for odds ratio 3 . 85: 1 . 29–11 . 46 ) ., None of the solitary Bordetella-free rabbits ( housed alone in isolation ) converted to disease-positive during this time ., Using the maternal transmission data ( Sampling Strategy Three in M&M for details ) , the importance of sibling-to-sibling versus mother-offspring routes of transmission was investigated ., First , the data revealed substantial correlation ( 0 . 53 ) among the infection fate of siblings and a highly significant litter-random effect ( litter variance =\u200a4 . 2±0 . 7 ) , demonstrating the importance of within-litter transmission ., Although the prevalence of B . bronchiseptica was significantly higher in does compared to kits ( Figure 4b; Z\u200a=\u200a5 . 03 p<0 . 0001 ) , having an infected mother did not significantly increase the probability of kits being infected ( infected mother: Z\u200a=\u200a1 . 74 , p\u200a=\u200a0 . 09 ) ., Nor was there any significant relationship between the litter random effect and the mothers prevalence status ( Z\u200a=\u200a−1 . 05×10−15 , p\u200a=\u200a1 . 0 ) ., We used MLST analysis to characterize the relationship between 90 isolates collected from four sampling efforts across rabbitry buildings ., All isolates were of sequence type ( ST ) -14 , which is a member of the B . bronchiseptica complex I 35 ., Thus , one circulating strain appears to dominate in our rabbit population ., This study demonstrates how FOI estimates coupled with random effects binomial regression analyses represent powerful tools for discerning between alternative modes of transmission for a directly transmitted pathogen ., Specifically , our results support a role for sibling-to-sibling transmission among young weaned kits as a major route of B . bronchiseptica spread in the rabbit population studied ( Figure 2 ) ., That the FOI reached a maximum value between 1 to 4 months of age – a time period when kits are re-housed in sibling pairs – followed by a sharp decline in the older age-classes , is consistent with high between-sibling transmission in young weanlings ( Figure 2 ) , regardless of host sex ( Figure 3 ) ., Results from the binomial regression analyses further support a major role for sibling-to-sibling transmission in driving B . bronchiseptica dynamics in the rabbitry; being co-housed with an infected sibling increased the risk of infection almost 4-fold ( Figure 4a ) ., In comparison , the data did not support all other potential transmission routes; namely maternal , breeder or environmental routes ., These insights shed light on the dynamics of disease spread and allow an assessment to be made of the best method ( s ) for effective long-term disease control , discussed more fully below ., A basic motivation for this study was to demonstrate how robust statistical tools can be used to disentangle routes and modes of transmission in humans and social animals from infection-at-age data ( within family groups ) , which is of broad medical , ecological and veterinary interest ., The FOI analyses we present may have greatest application for analyzing disease dynamics in medical and agricultural settings because here one often has direct access to date-of-birth information , knowledge of the distinct mixing patterns over the lifetime of the host , as well as host infection status ( for example , by detecting a serological response in the live animal , by the polymerase chain reaction ( PCR ) or by pathogen isolation ) ., One complexity which often arises in analyses of medical and agricultural diseases is clustering in the data; hosts live in families , litters or herds and once an infection is introduced , hosts within that cluster have a higher instantaneous rate of becoming infected than those outside the cluster ., Our use of random effect binomial regression analysis allows us to estimate the subject-specific measure of the effect 20 and evaluate the importance of social mixing in disease spread ., Thus , using the following protocol , the transmission dynamics of a range of directly transmitted infections can be analyzed by: ( 1 ) using the catalytic model and associated FOI analysis to determine the core susceptible age-class ( es ) ; ( 2 ) using random-effect binomial regression to inform on whether transmission is largely within the social group ( family/litter/herd etc ) or from an external social group; ( 3 ) carefully constructing transmission experiments , whenever possible , to test whether within-group versus between-group individuals are the dominant source of infection ., What might explain heterogeneities in rabbit susceptibility to B . bronchiseptica infection; for example , the decline in B . bronchiseptica prevalence in older-age classes ( in rabbits ∼20 months of age ) ?, Between-rabbit variation in protective anti-B ., bronchiseptica immunity – and hence resistance to infection – is likely to at least partly explain differences in host susceptibility to infection ., Indeed , recent work has shown that the protective immune response against B . bronchiseptica varies between individual rabbits , with robust serum IgG detected in some hosts for up to 5 months post infection , which correlated with clearance from the respiratory tract 28 ., Given the persistent nature of B . bronchiseptica infections in rabbits – infections of 5 months were routinely recorded 28 – and other mammals 23 , the decline in prevalence we observe is unlikely to be driven by bacteria clearance and recovery ., Rather , some level of enhanced immune protection in older age-classes may be responsible for conferring some level of anti-bordetella resistance ., Thus , the low attack rates ( or number of reported cases per unit time in a given age-class , divided by the number in that age class ) in older-age classes likely reflect low proportions of rabbits susceptible to infection – i . e . immune , disease-negative hosts – rather than a real decline in the rate at which susceptible rabbits acquire infection ., In addition , between-rabbit heterogeneities in protective anti- B . bronchiseptica immunity might also help explain differences in rabbit susceptibility to infection in the maternal- and co-housed sibling- transmission studies reported here ., Is there any epidemiological support for the major route of B . bronchiseptica spread ( sibling-to-sibling ) identified using our statistical framework ?, B . bronchiseptica is known to pass efficiently and spread rapidly between populations of young weaned pigs 40 , consistent with a sibling-to-sibling route for B . bronchiseptica transmission amongst young farmed animals ., This would be particularly true in agricultural systems where an all-in/all-out ( the facility is completely emptied and cleaned between groups of age-matched animals which move together between phases of production ) policy of animal breeding is not practised , as is the case in the rabbitry under study ., However , that our FOI estimates were above 0 . 1 before 1 month of age suggests some maternal or environmental transmission is occurring in young weanlings and may be key to initiating the sibling-to-sibling transmission which follows ., Indeed , a maternal route of transmission is thought initiate B . bronchiseptica infections in swine and rabbits 26 , 27 , but that infection only becomes endemic when passed horizontally between different batches of susceptible young 27 ., Interestingly , the time when FOI values peaked in young weanlings , coincided with a period where maternal protection wanes in kits – antibodies against B . bronchiseptica decreased between 2- 6 weeks of age in rabbits 25 – and could also contribute to increased susceptibility to infection observed in this age class ., Thus , based on our findings and the published literature , we propose that the cycle of B . bronchiseptica infection in our rabbitry is maintained by a proportion of chronically infected breeder females and males ( the infectious reservoir ) with the majority of transmission occurring between young weaned siblings ., One important application for the analytical tools presented here is in the implementation of targeted disease control programs ., Given that targeting those high-risk subgroups identified as playing key roles in transmission – rather than applying disease control measures randomly – is one efficient strategy to control disease 2 , 6 , a precautionary management approach might rely on the selective removal of infected weanlings to reduce sibling-to-sibling transmission ., Selective removal of breeder animals – which may represent potential maintenance hosts for B . bronchiseptica – may also improve disease control by eliminating the infectious reservoir ., Indeed , pre-emptive culling based on pre-determined patterns of disease spread has been successfully used to combat the spread of foot-and-mouth disease in cattle 41 , 42 ., The relationship between culling intensity and the resulting disease prevalence can be estimated when knowledge on population density and disease prevalence is available 43 ., This allows estimates to be made regarding the level of culling needed to produce significant reductions in disease prevalence ., The analyses presented here can be applied to a range of medical and veterinary diseases to better understand the dynamics and mechanisms of disease spread , provided they are directly transmitted and induce lifelong immunity to re-infection ., For example , the disease caused by mycobacterium – the etiological agent of tuberculosis in animals including bovine and humans – is largely directly transmitted , causes a sub-acute or chronic disease state which is irreversible 14 , 15 , 18 and can be routinely confirmed via culture , making it a tractable disease for application of FOI analyses ., Indeed , the tools of infectious disease quantitative epidemiology have successfully been applied to further understand Mycobacterium bovis infection dynamics in wildlife population of badgers 44 , ferrets 14 and bison 15 , 45 and has shed light on likely patterns of mycobacteria transmission in the wild ., However , these tools have not been used to the same effect in agricultural settings despite the debilitating effects of this disease and the potential to improve disease control therein ., Other veterinary diseases which are tractable for this type of analyses include brucellosis , bovine herpes infection , classical swine fever , bovine mastitis and atrophic rhinitis in swine , to name but a few ., Finally , the FOI model presented here can be extended to include diseases with reversion to non-diseased state or non-benign diseases ( i . e . associated with increasing death rate ) , or indeed to include a period where hosts are not exposed to infection ( for example , when maternal antibodies are known to provide protection against specific diseases early in life ) similar to a guarantee time in survival analysis ( see Caley & Hone 2002 for examples of such extensions ) ., Our study has some limitations ., Although the method we outline can clearly reveal the age-class for which most of the new infection occurs , it cannot easily discern whether that infection is mainly within an age-class versus from a different age-class ., However , once the high FOI age-class is identified , careful design of transmission experiments could confirm the likely source of infection , and such studies are underway in our University ., To control and possibly eradicate infectious diseases we need a better understanding of pathogen population dynamics and structure ., Indeed , only when HIV population structure was understood did the requirement for a three-cocktail HIV drug therapy become clear 46 ., Knowledge of pathogen population structure is also needed to determine which disease-associated genes are under directional selection change ., To this end we used MLST analysis to investigate whether strain-specific differences existed in the epidemiological pattern of infection 30 , 31 ., However , only one major circulating sequence type – ST14 – was identified in our rabbits regardless of rabbit age , sex or facility building ., The dominance of ST14 across our facility may be due to the successful expansion of this single serotype over time ., Alternatively , a limitation in sampling could have potentially biased our results; the sequence type of only 1 colony per swabbed plate ( i . e . per rabbit ) was determined at each sampling round ., Therefore if the rabbit was colonized with multiple strains we most likely detected the dominant type ( ST14 ) ., More intensive sequencing typing is required to test whether the lack of genetic variation we report is real and such studies are ongoing ., This study demonstrates the ease with which potential routes and reservoirs of infection can be discriminated amongst from age-prevalence data in medical , agricultural , and wildlife setting when we have access to fundamental age-prevalence data ., Much remains to be done to achieve a better understanding of the complex dynamics of chronic infections and to extend this model to incorporate factors such as host immunity and parasite genetic variation . | Introduction, Methods, Results, Discussion | Identifying the major routes of disease transmission and reservoirs of infection are needed to increase our understanding of disease dynamics and improve disease control ., Despite this , transmission events are rarely observed directly ., Here we had the unique opportunity to study natural transmission of Bordetella bronchiseptica – a directly transmitted respiratory pathogen with a wide mammalian host range , including sporadic infection of humans – within a commercial rabbitry to evaluate the relative effects of sex and age on the transmission dynamics therein ., We did this by developing an a priori set of hypotheses outlining how natural B . bronchiseptica infections may be transmitted between rabbits ., We discriminated between these hypotheses by using force-of-infection estimates coupled with random effects binomial regression analysis of B . bronchiseptica age-prevalence data from within our rabbit population ., Force-of-infection analysis allowed us to quantify the apparent prevalence of B . bronchiseptica while correcting for age structure ., To determine whether transmission is largely within social groups ( in this case litter ) , or from an external group , we used random-effect binomial regression to evaluate the importance of social mixing in disease spread ., Between these two approaches our results support young weanlings – as opposed to , for example , breeder or maternal cohorts – as the age cohort primarily responsible for B . bronchiseptica transmission ., Thus age-prevalence data , which is relatively easy to gather in clinical or agricultural settings , can be used to evaluate contact patterns and infer the likely age-cohort responsible for transmission of directly transmitted infections ., These insights shed light on the dynamics of disease spread and allow an assessment to be made of the best methods for effective long-term disease control . | A lack of understanding regarding determinants of infectious disease transmission has hindered improved disease control efforts ., Here we had the unique opportunity to study the natural transmission of the respiratory pathogen Bordetella bronchiseptica within a commercial rabbitry ., B . bronchiseptica is a directly transmitted gram-negative bacterium belonging to the genus Bordetella , which also comprises B . pertussis and B . parapertussis , the etiological agents of whooping cough in humans ., In this study we estimated the importance of rabbit sex , age and social group on disease spread ., To do this we first outlined a set of hypotheses about how natural B . bronchiseptica infections may be transmitted between rabbits ., We then discriminated between these hypotheses by estimating the rate at which susceptible individuals acquire infection ( or force-of-infection ) using B . bronchiseptica age-prevalence data ., The importance of social structure in disease spread was then evaluated using random-effect binomial regression ., Our results support young weanlings as the age cohort primarily responsible for B . bronchiseptica transmission and demonstrate that easy to collect age-prevalence data can be used to infer the likely age-cohort responsible for disease transmission ., Such insights shed light on the dynamics of disease spread and allow an assessment to be made of the best methods for effective disease control . | evolutionary biology/microbial evolution and genomics, ecology/evolutionary ecology, respiratory medicine/respiratory infections, public health and epidemiology/infectious diseases, infectious diseases/bacterial infections, ecology/population ecology | null |
journal.pbio.3000229 | 2,019 | Structural basis for neutralization of hepatitis A virus informs a rational design of highly potent inhibitors | Over the past 2 decades , progress in understanding human infections caused by hepatitis A virus ( HAV ) has been eclipsed by the priority of combating persistent hepatitis B virus ( HBV ) and hepatitis C virus ( HCV ) infections ., HAV , the most important agent for enterically transmitted viral hepatitis , is distributed worldwide and infects all age groups 1 ., The global burden of HAV has not abated ., Approximately 1 . 5 million clinical cases of HAV occur annually despite the availability of an effective vaccine 2 , 3 ., Hepatitis A as an infectious disease strongly correlates with income , hygiene , and living conditions 4 ., Areas with poor hygiene and living conditions continue to be under constant threat of HAV outbreaks 4 ., More recently , HAV has also started to become a new public health concern in well-developed , economically advanced countries due to the lack of natural or vaccine-induced acquired immunity to HAV in many adults 5 , 6 ., In the past year , more than 649 people throughout California have been reported to be infected with HAV ., Among these , 417 required hospitalization , and 21 patients died , making this the largest outbreak in the United States in the past 20 y 7 ., Development of antiviral therapy against HAV infection is urgently needed ., HAV , transmitted via the fecal–oral route , is a positive-sense , single-stranded RNA icosahedral virus belonging to the genus Hepatovirus within the Picornaviridae family 8 ., The 7 . 5 kb genome of HAV contains a single open reading frame ( ORF ) that encodes a giant polyprotein 9 ., The polyprotein is processed by a viral protease ( 3Cpro ) into 3 polypeptide intermediates , namely , P1–P3 9 ., P1 is subsequently further processed into 3 structural proteins , VP0 ( a precursor for VP2 and VP4 ) , VP3 , and VP1 , which self-assemble into a spherical capsid with icosahedral symmetry 10 ., Five copies of the VP1 capsid protein surround the icosahedral 5-fold axes ., Three copies of VP2 and VP3 alternate at the 3-fold axes , and 2 copies of VP2 abut each other at the 2-fold axes 11 ., Although a limited number of antigenic sites located on the HAV capsid have been revealed by escape mutants , the antigenicity of HAV is largely uncharacterized 12 , 13 ., Our recent study involving the structure of a complex of HAV with its neutralizing monoclonal antibody ( NAb ) , R10 , extended the previously unreported VP2 antigenic sites 14 ., Unlike other picornaviruses , HAV is extremely stable , both genetically and physically ., So far , 6 genotypes of human HAV have been identified 15 but with only a single serotype , suggesting that HAV has highly conserved antigenic sites 16 , 17 ., The low antigenic variation might be attributed to its highly deoptimized codon usage 18 ., A systematic and comprehensive study of the antigenic characteristics of HAV and neutralizing mechanisms could facilitate the design of effective small-molecule antivirals targeting HAV ., We set out to clarify the molecular basis for the antigenicity of HAV by characterizing 4 NAbs with varying neutralizing activities against the virus ., We sought this information for rationally designing antiviral inhibitors ., Here , we report the characterization of 4 highly potent NAbs: F4 , F6 , F7 , and F9 ., Furthermore , we have analyzed the experimentally derived high-resolution structures of HAV bound to the 4 NAbs as well as the previously reported R10-HAV structure to identify conserved epitopes for gaining key structure-activity correlates ., Using a robust in silico docking method , we have screened the DrugBank Database and have identified 1 promising inhibitor named golvatinib ., Cell-based antiviral assays have confirmed the ability of golvatinib to block infections caused by HAV ., To shed further light on the nature of the antigenicity of HAV , 2 rounds of monoclonal antibodies ( mAbs ) were generated ., R10 , an NAb with a 50% neutralization concentration value ( neut50 ) of approximately 2 nM 14 , was produced in the first round , and over 30 mAbs were screened during the second round ., Of these later antibodies , 4 antibodies , named F4 , F6 , F7 , and F9 , are NAbs ., Surface plasmon resonance ( SPR ) experiments showed that the 5 NAbs bind to HAV with a high affinity in the nanomolar range ( Fig 1A , S1 Fig ) ., A number of NAbs with similar affinities are virus specific ( e . g . , dengue virus-specific human mAb 5J7; Japanese encephalitis virus-specific mAbs 2F2 and 2H4 ) and exhibit exceptionally potent neutralizing activities 19 , 20 ., To investigate whether these 5 NAbs recognize different epitopes or the same patch of epitopes , we performed a competitive binding assay ., Briefly , the CM5 chip ( BIAcore , GE Healthcare ) , fully occupied with HAV , was initially saturated with R10 , and additional binding with another NAb was evaluated ., The CV60 mAb ( an mAb against Coxsackievirus A16 ) was used as a negative control ., Binding of R10 blocks the attachment of other 4 NAbs to HAV ( Fig 1B ) , suggesting that these 5 NAbs may bind to the same patch of epitopes or at least partially overlapped epitopes ., To characterize neutralizing activities , these 5 NAbs were evaluated for their abilities to prevent HAV infection ., Of note , all 4 NAbs generated from the second round showed potent neutralizing activities , of which F6 exhibited the strongest neutralizing activity ( a neut50 value of approximately 0 . 1 nM , which was 20-fold more potent than R10 Fig 1C ) ., To define precisely the atomic determinants of the interactions between these 4 NAbs and HAV , structural investigations of HAV in complex with fragment of antigen binding ( Fab ) from its NAbs were carried out ., Cryo-EM micrographs of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV complexes were recorded using a Titan Krios electron microscope ( Thermo Fisher ) equipped with a Gatan K2 detector ( Gatan , Pleasanton , CA ) ( S2 Fig ) ., The structures of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV were determined at resolutions of 3 . 90 , 3 . 68 , 3 . 05 , and 3 . 79 Å with 4 , 536 , 7 , 245 , 16 , 743 , and 3 , 798 particles , respectively , by single-particle techniques using the gold-standard Fourier shell correlation = 0 . 143 criterion 21 ( Fig 2A , Table 1 , S3 Fig and S4 Fig ) ., Densities attributable to residue backbones and side chains were recognizable in maps ( Fig 2A–2C ) ., These maps were of sufficient quality to allow the atomic modelling of most of the HAV capsid proteins and NAb Fab ., The structures of these 5 complexes are almost indistinguishable ., Differences are observed in the residues of the common complementary determining regions ( CDRs ) of NAbs ( r . m . s . d . for 12 , 473 Ca atoms less than 1 . 25 Å ) , which are consistent with the results of the competitive binding assays ( Fig 2A and S5 Fig ) ., There are 60 copies of NAb Fabs ( probably fully occupied ) bound to the virus in accordance with the level of electron density for the Fab ( Fig 2A ) ., Possibly correlated with its unusual stability , HAV capsid proteins exhibit no notable conformational changes upon binding to any NAbs ., Unlike EV71 or other picornaviruses in which several distinct patches for neutralizing epitopes have been reported 20 , 22 , 23 , 24 , all 5 NAb Fabs encircle edges of the pentameric building blocks of the virus , between the 2-fold and 3-fold axes ( Fig 2A and Fig 3A ) ., Examination of the possibility of binding of 2 arms of an immunoglobulin-G ( IgG ) molecule to the HAV surface showed that any 2 adjacent Fabs binding to the capsid could indeed mimic the 2 arms of a single IgG molecule ( S6 Fig ) ., Therefore , the IgG avidity for all 5 NAbs might be observed due to 2 Fab arms of an IgG on the surface of HAV being sufficiently close ., To explore the mechanism of neutralization , real-time reverse transcription PCR ( RT-PCR ) assays were performed to quantify the virus remaining on the cell surface , following exposure to antibodies’ previrus attachment to cells at 4°C ., The results reveal that these NAbs prevent HAV attachment to the permissive 2BS cell surface ( S7 Fig ) ., In summary , the high potencies of all 5 NAbs could be due to several reasons , including ( 1 ) higher avidity of the bivalent form of antibody , ( 2 ) the ability of the bivalent antibody to aggregate virus particles 14 , and ( 3 ) efficient block viral attachment to the host cell ., As expected , all Fabs exhibit a similar mode of binding , in which 1 Fab binds across the interface between the pentamers , interacting with VP2 and VP3′ from different pentamers ( Fig 2 ) ., The footprints of the 5 NAbs cover interaction areas ranging from approximately 970 Å2 to 1 , 290 Å2 , of which approximately 60% ( approximately 630 Å2 ) and approximately 40% ( approximately 440 Å2 ) are contributed by the heavy-chain and light-chain variable domains , respectively ( Fig 3A ) ., In line with this observation , F6 epitope contains more amino acid residues than other epitopes ( S1 , S2 , S3 , and S4 Tables ) , which is consistent with the results of binding affinities and neutralizing activities ( Fig 1 ) ., Given the fact that the F6 exhibits the most potent antiviral activity , the epitope analysis of these 5 NAbs is representative of the F6 mAb ., The heavy chain predominantly binds to the BC loop and EF loop of VP3 , whereas the light chain binds to the BC loop of VP2 and the BC loop of VP3 ( Fig 3B and S1 Table ) ., The epitopes on HAV capsid include residues S65 , R67 , and T71 in the BC loop and A198 and S201 of VP2; A68 , S69 , D70 , S71 , V72 , G73 , Q74 , Q75 , K77 , and V78 in the BC loop of VP3; and L141 , D143 , T145 , G146 , I147 , T148 , L149 , and K150 in the EF loop of VP3 ( Fig 3B and S1 Table ) ., The region of the F6 Fab that binds HAV comprises 4 of the 6 common CDRs: H1 ( residues 28–32 ) , H2 ( residues 52–57 ) , H3 ( residues 100–106 ) , and L1 ( residues 30–31 ) with , unusually , additional interactions contributed by the light-chain framework region ( L-FR; residues 45–55; Fig 3C and S1 Table ) ., The antibody components of these interactions include residues Y31 , R45 , Y48 , S51 , R52 , L53 , D55 , and Q59 from the light chain and residues N28 , Q30 , H31 , Y32 , Q52 , T53 , N54 , T56 , Y57 , R98 , N101 , I102 , E103 , C104 , H105 , and Y106 from the heavy chain ( Fig 3C ) ., Tight binding between the F6 fab and HAV capsid is facilitated by 33 hydrogen bonds and 9 salt bridges ( Fig 3C ) ., Structures of HAV in complex with 5 NAbs reveal that epitopes on HAV locate within the same patch and are extremely conserved ( Fig 4A ) , which is substantially different when compared to other picornaviruses , e . g . , at least 4 regions of the epitopes recognized by its NAbs are mapped in EV71 26 , 27 ., In line with neutralizing activities , F6 epitope has 3 extra residues ( D70 , K77 , and L141 of VP3 ) when compared to others , and R10 possesses the least number of epitope residues ( all the NAbs recognized the A198 of VP2 except R10 ) ( Fig 4A ) ., These 5 NAbs share high sequence similarities at the framework region but bear relatively low sequence identities ( approximately 35% ) at the CDRs ( Fig 4B ) ., In spite of variations in the sequences , these 5 CDRs involved in the interactions with HAV adopt an indistinguishable configuration and a similar binding mode ( Fig 2 , Fig 4B , and S4 Fig ) ., As expected , further tight binding of F6 and F4 to HAV is made possible by the additional hydrogen bonds and charge interactions formed by the antibodies ( Fig 4C ) ., Furthermore , residues in R10 that interact with HAV are also fewer in number than those observed for other NAbs ., To decipher the structure activity correlates between HAV-NAb interactions and neutralizing activities , the interaction interface areas and binding energies were calculated and then compared with their neutralizing activities ( S5 Table ) ., We assembled a data set of 5 NAbs inhibition data for HAV and generated correlation plots between the Neut50 values and the area and energy of interaction , which produced a compelling correlation of 0 . 93 and 0 . 94 , respectively ( Fig 4D–4E ) ., These analyses suggest two lessons: ( 1 ) epitopes revealed by NAbs on HAV are good targets for drug design; and ( 2 ) the more robust binding of NAbs to the epitopes , the better the antiviral activities ., To date , 6 genotypes of human HAV have been identified but with only a single serotype ., This indicates that these 5 NAbs are likely to bind strongly to the 6 human HAVs and could be capable of preventing human HAV infections ., Sequence and structural analyses show that the residues constituting the epitopes are 87 . 5% identical and 94 . 6% conserved , with only 3 out of 21 contacting residues ( residue 67 of VP2 , residues 145 and 146 of VP3 ) being moderately conserved ( 50%–85% ) , and the remaining residues completely conserved ( 100% ) ( Fig 5A and Fig 5B ) ., The variation rate for the epitopes is even slightly lower than that for the whole capsid ( P1 , approximately 94 . 3% conserved ) , highlighting a single , conserved antigenic site for HAV ., Given the fact that a single , conserved antigenic site exists in HAV and the key structure-activity correlates based on the antigenic site have been established , we next used the structural data to rationally design and screen potent compounds against HAV targeting the antigenic site ., These residues composing the antigenic site are distributed on both sides of a long “gully , ” which forms a potential inhibitor binding pocket ( Fig 6A ) ., Results of our previous study have also indicated that the “gully” area might be critical for HAV receptor binding 14 ., We postulated , on the basis of inspection of the HAV-NAbs binding interface , that a tight binder ( a compound ) mimicking the NAbs might efficiently block HAV entry and infection ., To test this hypothesis , we scanned in silico the DrugBank database ( https://www . drugbank . ca/ ) , using Phase version 3 . 7 30 , Glide version 6 . 1 31 to identify potential tight binders ., Briefly , the 4 key residues 31 , 32 , 101 , and 102 from the heavy chain of the 5 NAbs , which made the greatest contributions to the specific action of antigen–antibody , were selected as a reference structure for pharmacophore modeling ., The generated pharmacophore was used to screen the drugs database of DrugBank ., A total of 2 , 588 drugs were screened ., Then , all selected drugs were docked to the antigenic site , and the top-ranked 4 molecules were selected ., Distinct from the others , compound 3 has the best glide score ( S6 Table ) ., Therefore , compound 3 , named golvatinib ( DB11977 ) , was predicted to bind to the “gully” much stronger than others ( Fig 6B ) ., As expected , in this docking pose , golvatinib contacts with the epitope residues , including S65 from VP2 and V72 , G73 , Q74 , Q75 , V78 , P79 , T144 , T148 , L149 , and Q246 from VP3 , via hydrophilic and hydrophobic interactions ( Fig 6A ) ., We therefore also measured the inhibitory activities of golvatinib by in vitro studies in 2BS cells ., We used 100 50% tissue culture infective dose ( TCID50 ) virus in the presence of different concentrations of the compounds and exposed control wells to the equivalent concentration of solvent ( DMSO ) to ensure no effects on uninfected cells or on virus titer ( S8 Fig ) ., The compound golvatinib exhibited a potent antiviral activity , with a 50% inhibitory concentration ( IC50 ) of approximately 1 μM , inhibiting the viral titer to below 15% at concentrations over 8 μM ( Fig 6B ) ., Meanwhile , no notable cytotoxic effect of golvatinib at concentrations of 0 . 0005 to8 μM was observed ( Fig 6B ) ., The measured antiviral activity of golvatinib is in agreement with that predicted in silico ., As expected , golvatinib , like the 5 NAbs , inhibits HAV infection by blocking attachment to the host cell ( S9 Fig ) ., Due to the partial overlapped binding sites of the NAbs and of golvatinib ( Fig 6A ) , it is quite possible that they are capable of competing each other to attach the HAV surface ., In addition , the binding of golvatinib to the HAV does not alter its particle stability ( S9 Fig ) , which is consistent with our previous results that stabilization or destabilization is unlikely to be the major neutralization mechanism in our study systems 14 ., Attachment of the virus to its cellular receptors located on the surface of the host cell and uncoating of the virus leading to the release of the viral genome into host cells are regarded as the 2 key steps for the successful entry of nonenveloped viruses , including picornaviruses , into host cells 32 ., Neutralizing antibodies block the entry of viruses into host cells by blocking the attachment of the virus to the cellular receptor 33 , overstabilizing the virus 34 , preventing the release of viral genome 27 , or physically destabilizing the capsid of the virus 22 ., In our previous study , we demonstrated that R10 , a HAV-specific neutralizing antibody , neutralizes HAV infection by preventing the binding of the virus to its putative receptor T-cell immunoglobulin and mucin-containing domain 1 ( TIM-1 ) 14 ., Recent evidences suggests that TIM-1 is not an essential receptor for the naked ( unenveloped ) HAV but rather an attachment factor for quasi-enveloped virions 35 , making the bona fide receptor ( s ) elusive ., Therefore , it is challenging to verify whether the binding of NAbs or golvatinib blocks the interactions between HAV and its bona fide receptor ., Many picornaviruses use cell-surface molecules belonging to the immunoglobulin superfamily ( IgSF ) as their cellular receptors , which usually consist of tandem repeats of between 1 and 5 Ig-like domains to interact with viruses 36 ., Given the fact that R10 competitively blocked TIM-1 Ig V binding to HAV 14 , it is possible that the bona fide receptor might be from the IgSF ., In this study , the antigen binding site of the newly screened NAbs ( F4 , F6 , F7 , and F9 ) maps to the same epitope on the surface of HAV as that identified for R10 , suggesting that the binding sites of the 4 NAbs and the bona fide receptor may overlap ., Previous studies have shown that residues S102 , V171 , A176 , and K221 of VP1 and D70 , S71 , Q74 , and 102–121 of VP3 are part of the neutralizing epitopes 13 ., However , structural analysis reveals that these putative epitope residues are forming 2 clusters that are separated by a distance of 40 to 50 Å on the HAV surface , suggesting that these residues are unlikely to form a single antigenic site ., The high-resolution structures of HAV in complex with 4 NAbs described in this study coupled with the results of our previous studies on HAV-antibody complexes 14 further verify the fact that VP2 ( but not VP1 ) , as well as VP3 , form a single , conserved antigenic site on the surface of HAV , which differs radically from the architecture of the antigenic sites of other picornaviruses 37 , 38 ., However , our studies cannot exclude the possibility of the likely existence of a second neutralizing antigenic site involving residues in VP1 , yet ill-defined on the viral capsid ., Additionally , the neutralizing epitopes should differ with those of binding but non-neutralizing antibodies , which needs to be further investigated ., The single , conserved antigenic site we identify could serve as an excellent target for structure-based drug design ., Although hepatitis A is a vaccine-preventable disease 39 , an anti-HAV drug would be indispensable for treating fulminating infections ., In this study , about 2 , 588 drug candidates ( compounds ) from the DrugBank database were selected for in silico docking studies ., One of the candidates predicted to interact with the conserved antigenic site by the docking studies exhibited excellent antiviral activity without any notable cytotoxicity ., Therefore , based on the preclinical evaluation of its cytotoxicity and pharmacodynamics , golvatinib , previously investigated for the treatment of platinum-resistant squamous cell carcinoma of the head and neck 40 , could act as a lead compound for anti-HAV drug development ., In summary , we have used a combined experimental and computational approach starting from a number of NAbs targeting a single , conserved antigenic site located on the surface of a complete viral capsid to obtain , in a single round of design , a potent micromolar-range drug candidate that is effective and safe and has many drug-like properties ., Animals were bred and maintained under specific pathogen-free ( SPF ) conditions in the institutional animal facility of the Institute of Biophysics , Chinese Academy of Sciences ., All animal experiments were performed with protocols ( protocol numbers VET102 , VET201 , VET203 , and VET301 ) approved by the Animal Care and Use Committee of Institute of Biophysics , Chinese Academy of Sciences ., HAV virus genotype TZ84 ( HAV IA genotype ) was used to infect 2BS cells at a multiplicity of infection ( MOI ) of 0 . 2 at 34°C ., Particle production and purification have been described previously 11 ., F4 , F6 , F7 , and F9 were purified from mouse ascites with a protein A affinity column ( GE ) ., The Fab fragment was generated using a Pierce FAB preparation Kit ( Thermo Scientific ) , according to the manufacturer’s instructions ., Briefly , after removal of the salt using a desalting column , the antibody was mixed with papain and then digested at 37°C for 6 h ., The Fab was separated from the Fc fragment by using a protein A affinity column ., Then , Fab was loaded onto a Hitrap Q FT column ( GE ) ., Fractions corresponding to the major peak were collected and concentrated for cryo-EM analysis ., The binding affinities of the 5 NAb assays were determined by SPR ., These experiments were performed using a BIAcore 3 , 000 machine ( BIAcore , GE Healthcare ) in the buffer solution containing 10 mM HEPES ( pH 7 . 4 ) , 150 mM NaCl , and 0 . 005% v/v Tween 20 at 25°C ., The purified HAV full particles were directly immobilized onto CM5 sensor chips ( BIAcore , GE Healthcare ) at concentrations equivalent to approximately 950 response units ( 0 . 3 mg/ml ) ., Subsequently , gradient concentrations ( 0 . 0315 , 0 . 0625 , 0 . 125 , and 0 . 25 μM ) of purified Fab fragments of F4 , F6 , F7 , F9 , and R10 were used to flow over the chip surface ., To regenerate the chip , 100mM NaOH was used ., The binding affinities were analyzed using steady state affinity with the software BIAevaluation version 4 . 1 ., Binding competition between HAV antibodies was determined using SPR ( BIAcore 3 , 000 , GE ) ., The entire experiment was performed at 25°C in the buffer solution containing 10 mM HEPES ( pH 7 . 4 ) , 150 mM NaCl , and 0 . 005% v/v Tween 20 ., The CM5 biosensor chip ( BIAcore , GE Healthcare ) immobilized with HAV full particles ( 0 . 3 mg/ml ) was first saturated with R10 for 5 min ., Afterward , the other NAbs were injected in the presence of R10 for another 3 min ., CV60 , an irrelevant antibody , was used as a negative control ., Except for R10 , all other NAbs were evaluated at a concentration of 300 nM for saturation ., R10 was applied at a concentration of 900 nM ., The chip was regenerated with 100mM NaOH ( GE Healthcare ) ., For the neutralization assay , purified mAbs at a concentration of 0 . 2 mg/ml were initially diluted 8-fold as stocks and then serially diluted 2-fold with DMEM containing 2% FBS; 100 μl of 2-fold antibody dilutions were mixed with 100 μl of HAV containing 100 TCID50 for 1 h at 37°C and then added to monolayers of 2BS cells in cell culture flasks ( T25 CM2 ) ., Meanwhile , maintaining medium was provided as well ., Each dilution was replicated 3 times along with one control that contained no antibody dilution ., After 21 d of growth at 34°C , the medium was removed , and the cells were washed three times using PBS buffer; 1 ml of Trypsin/EDTA was added , and the flask was left for 3 min at 37°C ., The suspended cells were freeze-thawed 5 times to collect the virus ., Enzyme-linked immunosorbent assay ( ELISA ) was used to measure HAV antigen content ., The percent inhibition was determined relative to the mean OD450 values of the control wells in which the virus has been incubated with medium alone ., Purified F4 , F6 , F7 , and F9 Fab fragments were incubated with purified HAV ( at a concentration of 2 mg/ml ) separately on ice for 10 min at a ratio of 120 Fab molecules per virion ., A 3-μl aliquot of the mixtures of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fb-HAV were transferred onto a freshly glow-discharged 400-mesh holey carbon-coated copper grid ( C-flat , CF-2/1-2C; Protochips ) ., Grids were blotted for 3 . 5 s in 100% relative humidity for plunge-freezing ( Vitrobot; FEI ) in liquid ethane ., Cryo-EM data sets were collected at 300 kV using a Titan Krios microscope equipped ( Thermo Fisher ) with a K2 detector ( Gatan , Pleasanton , CA ) ., Movies ( 25 frames , each 0 . 2 s , total dose 30 e− Å−2 ) were recorded with a defocus of between 1 and 2 . 5 μm using SerialEM 41 , which yields a final pixel size of 1 . 35 Å ., The frames from each movie were aligned and averaged for the correction of beam-induced drift using MOTIONCORR 42 ., Particles from micrographs were picked automatically using ETHAN 43 and then manually screened using the boxer program in EMAN 44 ., The CTF parameters for each micrograph were estimated by using a GPU accelerated program Gctf 45 ., Cryo-EM structures were determined with Relion 1 . 4 46 with the application of icosahedral symmetry ., The initial model was created by EMAN2 47 ., A total of 4 , 536 , 7 , 245 , 16 , 743 , and 3 , 798 particles of F4-Fab-HAV , F6-Fab-HAV , F7-Fab-HAV , and F9-Fab-HAV were used to determine structures at resolutions of 3 . 9 , 3 . 68 , 3 . 05 , and 3 . 79 Å , respectively , as evaluated by the so-called gold standard FSC procedure between 2 half maps ( threshold = 0 . 143 ) 21 ., The crystal structure of HAV full particle ( PDB ID code 4QPI ) was used to fit the complex EM maps , and the atomic models of F4- , F6- , F7- , and F9-Fabs were built de novo into densities with the structure of R10 ( PDB ID code 5WTG ) as a guide , using COOT 48 ., All models were further refined by positional and B-factor refinement in real space using Phenix 49 and rebuilding in COOT 48 iteratively ., The final models were evaluated by Molprobity 50 functions integrated in Phenix ., Data and refinement statistics are summarized in Table 1 ., In antigen binding systems , the chain C , D , E , and G of F4 Fab-HAV , F6 Fab-HAV , F7 Fab-HAV , and F9 Fab-HAV structures were fetched out to perform MD simulation and binding energy calculations ., In the 4 chains , the chain D and E were set to ligand and the chain C and G were set to receptor ., The complex was solvated to TIP3P waters , and 0 . 1 M NaCl was added to systems as salt with soft tleap in AmberTools 16 51 ., Amber 16 was used to perform MD simulation ., All 4 systems were first relaxed by 5 , 000-step minimization ( 2 , 000 steps , steepest descent minimizations; 3 , 000 steps , conjugate gradient minimization ) ., After minimization , the system was gradually heated from 0 K to 300 K in the canonical NVT ensemble with a Langevin thermostat using a collision frequency of 2 . 0 ps−1 ., Initial velocities were assigned from a Maxwellian distribution at the starting temperature ., Then 100 ps of density equilibration with weak restraints on the complex was followed by 500 ps of constant pressure equilibration at 300 K and 1 atm ., Finally , 10 ns MD simulations for each system was conducted with the target temperature at 300 K and the target pressure at 1 . 0 atm ., In 4 systems , Na+ was selected as the counter ions; the concentration of NaCl was set to 0 . 1 M , and ions’ parameters of Joung and Cheatham 52 were used ., Electrostatics was handled using the particle mesh Ewald ( PME ) algorithm 53 with a 10 . 0 Å direct−space nonbonded cutoff ., All bonds involving hydrogen atoms were constrained using the SHAKE algorithm 54 , using a time step of 2 . 0 fs ., The coordinates’ trajectories were saved every 2 ps during the whole MD runs ., MM-GBSA 55 was used to calculate the binding energy of F4 , F6 , F7 , F9 , and R10 Fab ( chain D and chain E ) and HAV VP2 and VP3 ( chain C and chain G ) ., In each system , 100 snapshots of the last 6 ns MD simulation were fetched out to calculate the binding energy ., The entropy contributions were neglected because the same receptor was used and because the normal mode analysis calculations are computationally expensive and subject to a large margin of error that introduces significant uncertainty in the result ., The free energy for each species ( ligand , receptor , and complex ) is decomposed into a gas-phase MM energy , polar , and nonpolar solvation terms , as well as an entropy term , as shown in the following equation:, ΔG=ΔEMM+ΔGsolv−T∙ΔS=ΔEbat+ΔEvdw+ΔEcoul+ΔGsolv . p+ΔGsolv . np−T∙ΔS ., EMM is composed of Ebat ( the sum of bond , angle , and torsion terms in the force field ) , a van der Waals term , EvdW , and a Coulombic term , Ecoul ., Gsolvp is the polar contribution to the solvation free energy , often computed via the Generalized-Born ( GB ) approximation ., Gsolvnp is the nonpolar solvation free energy , usually computed as a linear function of the solvent-accessible surface area ( SASA ) ., The Phase program of the Schrodinger Suite 2013 30 was used for the pharmacophore modeling ., Four key residues at positions of 31 , 32 , 101 , and 102 from the heavy chain of the 5 NAbs were selected as a reference structure for modeling ., Pharmacophore sites were generated using the default set of chemical features: hydrogen bond acceptor ( A ) , hydrogen bond donor ( D ) , hydrophobe ( H ) , negative ionizable ( N ) , positive ionizable ( P ) , and aromatic ring ( R ) ., The size of the pharmacophore box was set to 1 Å to optimize the number of final common pharmacophore hypotheses ., The generated pharmacophore was used to screen the drugs database of Drugbank ., The distance matching tolerance was set to 2 . 0 Å ., A total of 2 , 588 drugs were screened out using this procedure ., The docking algorithm Glide 31 , which is based on descriptor matching , was used to perform virtual screening and learn the interactions between small molecules and the protein structure ., The structure of HAV epitopes was prepared and then used to build the energy grid ., For Glide docking , the docking box was centered on the position of mass center of the 4 selected residues , and its outer box size was set to 40 × 40 × 40 Å ., The scaling factor for protein van der Waals radii was set to 1 . 0 ., All 2 , 588 drugs were docked to the antigen binding site on HAV capsids , and the 4 molecules were selected ., The compound 3 , which has the best glide score , was selected finally ., Approximately 2 × 105 2BS cells were seeded into each well of a 24-well plate and incubated overnight in a CO2 incubator supplemented with 5% CO2 ., Before virus infection , HAV ( 100 TCID50 ) was incubated with serially diluted concentrations ( 0 , 0 . 008 , 0 . 032 , 0 . 125 , 0 . 5 , 2 , and 8 μM ) of golvatinib ( MedChemExpress ) for 1 h at room temperature with gentle rocking and then transferred to the plate containing 2BS cells ., After adsorption for 1 h , the inoculum was removed , and the cells were supplied with fresh maintenance medium and incubated at 34°C ., At 7 d post infection , the cells were lysed for ELISA to determine HAV antigen content ., Approximately 2 × 105 2BS cells were seeded into 24-well plates and incubated overnight in a CO2 incubator ., Inhibitors were serially diluted concentrations ( 0 , 0 . 008 , 0 . 032 , 0 . 125 , 0 . 5 , 2 , and 8 μM ) and then transferred to the plate containing 2BS cells ., Seven days after addition of drug , CCK-8 kit ( Sangon Biotech ) was used according to the manufacturer’s protocol ., In brief , each well of the plate had10 μl CCK-8 solution added and was incubated 2 h at 37°C ., | Introduction, Results, Discussion, Materials and methods | Hepatitis A virus ( HAV ) , an enigmatic and ancient pathogen , is a major causative agent of acute viral hepatitis worldwide ., Although there are effective vaccines , antivirals against HAV infection are still required , especially during fulminant hepatitis outbreaks ., A more in-depth understanding of the antigenic characteristics of HAV and the mechanisms of neutralization could aid in the development of rationally designed antiviral drugs targeting HAV ., In this paper , 4 new antibodies—F4 , F6 , F7 , and F9—are reported that potently neutralize HAV at 50% neutralizing concentration values ( neut50 ) ranging from 0 . 1 nM to 0 . 85 nM ., High-resolution cryo-electron microscopy ( cryo-EM ) structures of HAV bound to F4 , F6 , F7 , and F9 , together with results of our previous studies on R10 fragment of antigen binding ( Fab ) -HAV complex , shed light on the locations and nature of the epitopes recognized by the 5 neutralizing monoclonal antibodies ( NAbs ) ., All the epitopes locate within the same patch and are highly conserved ., The key structure-activity correlates based on the antigenic sites have been established ., Based on the structural data of the single conserved antigenic site and key structure-activity correlates , one promising drug candidate named golvatinib was identified by in silico docking studies ., Cell-based antiviral assays confirmed that golvatinib is capable of blocking HAV infection effectively with a 50% inhibitory concentration ( IC50 ) of approximately 1 μM ., These results suggest that the single conserved antigenic site from complete HAV capsid is a good antiviral target and that golvatinib could function as a lead compound for anti-HAV drug development . | Hepatitis A virus ( HAV ) is a unique , hepatotropic human picornavirus that infects approximately 1 . 5 million people annually and continues to cause mortality despite a successful vaccine ., There are no licensed therapeutic drugs to date ., Better knowledge of HAV antigenic features and neutralizing mechanisms will facilitate the development of HAV-targeting antiviral drugs ., In this study , we report 4 potent HAV-specific neutralizing monoclonal antibodies ( NAbs ) , together with our previous reported R10 , that efficiently inhibit HAV infection by blocking attachment to the host cell ., All 5 epitopes are located within the same patch and are highly conserved across 6 genotypes of human HAV , which suggests a single antigenic site for HAV , highlighting a prime target for structure-based drug design ., Analysis of complexes with the 5 NAbs with varying neutralizing activities pinpointed key structure-activity correlates ., By using a robust in silico docking method , one promising inhibitor named golvatinib was successfully identified from the DrugBank Database ., In vitro assays confirmed its ability to block viral infection and revealed its neutralizing mechanism ., Our approach could be useful in the design of effective drugs for picornavirus infections . | antimicrobials, medicine and health sciences, immune physiology, enzyme-linked immunoassays, pathology and laboratory medicine, pathogens, drugs, immunology, microbiology, electron cryo-microscopy, viruses, microscopy, pharmacology, antibodies, immunologic techniques, research and analysis methods, monoclonal antibodies, immune system proteins, proteins, medical microbiology, antigens, microbial pathogens, immunoassays, hepatitis viruses, viral packaging, viral replication, hepatitis a virus, biochemistry, virology, viral pathogens, physiology, electron microscopy, microbial control, biology and life sciences, antivirals, organisms | Structures of hepatitis A virus in complex with five neutralizing antibodies reveal a single conserved antigenic site and pinpoint key structure-activity correlates, allowing in silico screening to identify a potent candidate inhibitor drug, golvatinib. |
journal.pcbi.1003943 | 2,014 | Multitask Learning of Signaling and Regulatory Networks with Application to Studying Human Response to Flu | The relative ease of high-throughput data collection enables profiling a system of interest in many ways with complementary assays , at different times , and under various perturbations to compare and contrast the outcomes ., The resulting computational challenge is to develop analysis strategies that maximally leverage these related experiments to improve our ability to reconstruct biologically accurate models ., Even when applied to study the same condition , different types of high-throughput data ( e . g . , functional genetic screens and gene expression ) often times implicate largely disjoint groups of genes or proteins because each experiment highlights different facets of the biological processes and networks involved 1 ., Consequently , there has been extensive research to develop techniques for integrating one or more types of condition-specific high-throughput data with general purpose physical interaction networks , such as protein-protein interactions ( PPIs ) , to reconstruct signaling and regulatory networks 1–3 ( see 4 for a review ) ., These methods discern how the genes identified in complementary types of experiments relate to one another in a network context and propose new condition-specific regulators that are not directly observed to be relevant in the original data but form connections in the inferred networks ., Due to the dynamic nature of biological systems , especially those controlling stimulus response and development , it is critical to observe genome-wide changes over time 5 ., As reviewed in 5 , there are now computational approaches that exploit the unique structure in temporal datasets ( e . g . , time series gene expression ) to model dynamic processes and reverse engineer regulatory networks 6 , 7 ., Recent algorithms integrate temporal data and PPI networks to improve signaling pathway prediction by capitalizing on the dynamic information 8 , 9 ., Despite advances in modeling the temporal dimension and different types of assays per condition , there has been considerably less progress made for datasets that contain multiple related perturbations or stimuli ., Typically each condition is analyzed in isolation , and a post-processing comparison of the independent models is required to draw conclusions across conditions 9 , 10 ., Individual models of related conditions are required to appreciate the unique aspects of each , but building these models independently ignores that the observations may be generated from structurally similar networks ., As an example , consider the case of host gene expression following virus infection ., Although different viruses do not have identical effects on the host ( hence the gene expression patterns are unique to each virus ) , they also commonly affect a similar core set of host proteins ., These include Toll-like receptors ( TLRs ) , which recognize a large number of RNA viruses and activate a downstream pathway that leads to common expression response 11 , 12 , and other elements of innate immune response pathways 13 ., Similarly , in yeast several different types of stresses activate a large common set of genes ( termed the environmental stress response genes 14 ) , and additional examples abound in other species ., When modeling such responses , one may be able to take advantage of these commonalties without sacrificing the ability to reconstruct individual models for each response ., This type of machine learning is termed Multi-task learning 15 and usually applies to cases where one learns models for different problems that share information and/or parameters ., A key advantage of such framework is the ability to utilize additional data from related conditions when reconstructing networks for a specific response ., This is especially important when reconstructing biological response networks from high-throughput data because the number of parameters to fit is very large relative to the number of samples ., In addition , extensive data from a well-characterized condition may be able to compensate for sparse data in a similar , less-understood condition ., Multi-task learning has been applied to other problems in the biological domain including classification 16 , genome-wide association studies 17 , 18 , protein structure 19 , and pairwise protein-protein interaction prediction 20 , 21 ., Multi-commodity flow 22 and iterative applications of a prize-collecting Steiner forest algorithm 23 have been used to simultaneously reconstruct related response or disease networks , but these methods do not employ multi-task learning ., In addition , these previous approaches operate on static data and cannot account for the dynamic behaviors that are crucial for understanding many types of stimulus responses ., Here we present the Multi-Task Signaling and Dynamic Regulatory Events Miner ( MT-SDREM ) , which uses multi-task learning to reconstruct response pathways and temporal regulatory networks ., MT-SDREM is equipped to capitalize on the many dimensions in complex systems biology datasets by integrating different types of experimental data in each condition , explaining the time-dependent elements of a response ( as observed in gene expression data ) , and constraining the inferred networks to be similar for related conditions or perturbations ., Like its single-condition predecessor 8 , MT-SDREM iterates between finding pathways that connect the upstream proteins that directly interact with an external stimulus ( called source proteins ) and the downstream transcription factors ( TFs ) that regulate the response and learning dynamic regulatory networks activated by these TFs ., The learning process involves the simultaneous reconstruction of several such networks ., While a different network is learned for each condition , the joint learning framework allows sharing and/or constraining parameters across the different networks which helps overcome the overfitting problem that is often an issue when reconstructing biological networks ., We demonstrate how MT-SDREM can be used to gain insights into a clinically-relevant problem: characterizing the human response to viral infection ., In particular , we explore the differences between mild , seasonal strains of the influenza A virus , which are typically H1N1 or H3N2 strains 24 , and lethal , pandemic strains such as the H1N1 1918 Spanish flu and highly pathogenic avian H5N1 strains ., Influenza A strains are subtyped and named by their hemagglutinin ( HA ) and neuraminidase ( NA ) proteins ., Although there are presently 18 known HA subtypes and 11 NA subtypes 25 only a fraction of these have have infected humans ., Previous studies have characterized some of the differences between seasonal and pathogenic strains ., Seasonal H1N1 and H3N2 and highly pathogenic H5N1 influenza strains infect macrophages at similar rates , but H3N2 and H5N1 causes apoptosis more rapidly than H1N1 24 ., H1N1 also lead to weaker induction of MAPK signaling pathways than the H3N2 and H5N1 strains 24 ., Genomic comparisons of human and avian influenza strains identified 52 species-associated positions that could potentially enable an avian strain to cross over to humans if mutated 26 ., Influenza strains also vary in the cells and tissues they infect 27 , 28 with highly-virulent strains causing more widespread inflammation , including in the alveoli 27 ., Highly pathogenic strains have been shown to induce a stronger inflammatory cytokine response than seasonal influenzas 28 and the host inflammatory response is often more deadly during infection than the pathogen itself 29 ., However , much remains unknown about the host factors that are required for viral replication or to mount cellular defenses ., We study three strains of the influenza A virus — seasonal H1N1 , seasonal H3N2 , and highly pathogenic avian H5N1 — to explain how common host proteins react to the viral infection in a similar manner despite the differences in the temporal transcriptional programs that are activated ., The MT-SDREM networks identified many known regulators of influenza response and also suggested putative novel regulators ., Because the responses are jointly modeled using the multi-task setting , MT-SDREM is able to correctly recover TFs that are important drivers of the immune response that are missed when each viral strain is analyzed independently 10 and by previous methods for combining gene expression data across experiments ., In addition , MT-SDREM networks are more predictive of host genes that are required for viral replication , a potentially clinically-relevant phenotype 29 , than corresponding independent models or gene prioritization algorithms ., To test the advantages of multi-task learning we compared MT-SDREM with previous methods that can be used to analyze expression and interaction data ., Since we are not aware of prior methods that utilize multi-task learning in biological network reconstruction we first looked at the differences between applying MT-SDREM and applying SDREM separately to each of the three flu datasets ., We have also compared MT-SDREMs results to a baseline joint ranking of differentially expressed ( DE ) genes from different experiments in a single analysis ., This approach is similar to several previous studies that perform follow up analysis using such joint sets 52 ., Since the ground truth ( complete underlying networks for each condition ) is obviously unknown , we used three different types of complementary information for these comparisons ., First , we examined the set of TFs identified by each of these methods and determined their relevance to the condition being studied ., Next , we used the Gene Ontology ( GO ) to test the differences in the identified functional categories between the different analysis methods ., Expression experiments and RNA interference ( RNAi ) screens have revealed a multitude of host pathways and processes that are involved in viral host response including MAPK signaling , apoptosis , trafficking , mRNA export , splicing , and proteolysis 30 , 53 , 54 ., A statistical meta-analysis implicates nearly 3000 host genes 55 in these pathways ., Although many processes as a whole are relevant to influenza response , not all genes participating in those processes necessarily are important ., Therefore we focused our TF and GO evaluation on immune processes , which were shown to compose a critical component of the host response that kills infected cells , protects uninfected cells , combats viral components , and promotes inflammation 56 Finally , we used a set of RNAi experiments that were performed for H1N1 and H5N1 to test the ability of these different methods to identify key disease related proteins ., In these experiments proteins are knocked down one at a time and the impact on viral load is measured ., A protein affecting viral load is likely participating in the host response and so methods that can identify such proteins more accurately are in better agreement with the observed response ., The RNAi data for H1N1 was obtained from 30 , 53 , 54 , 57 , 58 resulting in a total of 980 screen hits , 925 of which were present in our initial interaction network ( which contained 16671 genes , Methods ) ., 32 screen hits for H5N1 were obtained from 57 , all of which are present in our interaction network ., In Table 1 we present the overall and pairwise overlap of the inferred TFs for the 3 conditions ( extracted by same mechanism as in SDREM 8 , 10 ) for MT-SDREM and compare it to when SDREM is run independently on the 3 conditions ( I-SDREM ) ., Note that the pairwise intersections shown are in addition to the overall intersection between all of the 3 conditions ., The shared TFs identified by MT-SDREM among all 3 conditions that are missed by I-SDREM include several that are known to be immune response related ., In particular , CEBPA is known to be responsible for regulating a large variety of cell functions including immune and inflammatory response 59 ., MT-SDREM also identifies SMAD4 in all three conditions ., SMAD family proteins are part of the TGF pathway as mentioned above ., MT-SDREM also identifies RB1 which has been implicated in viral immune response 60 , JUN which is part of the AP-1 TF complex , and PPARG an important TF regulating immune response mentioned above ., In contrast , I-SDREM does not identify any TF in the intersection that MT-SDREM does not ., In addition , we also find several immune response related TFs in the pairwise overlaps for MT-SDREM that we do not see for I-SDREM ., For the overlap between H1N1 and H3N2 , MT-SDREM identifies IRF1/3/5 which are known to regulate interferons and thus important for immune response ., For the overlap between H1N1 and H5N1 , MT-SDREM finds the the STAT3 gene which is part of the JAK-STAT signaling pathway and ATF2 , part of the AP-1 TF complex ., For the pairwise intersection of H1N2 and H3N2 , I-SDREM identifies NR3C1 as a TF while MT-SDREM only selects it as an intermediate ( signaling ) protein ., It also identifies another member of the SMAD family ( SMAD3 whereas MT-SDREM identifies SMAD4 ) ., For H3N2 and H5N1 it identifies AHR whose activation inhibits inflammation 61 and RELA in the intersection of H1N1 and H5N1 , which as part of the NF- complex ., We also compared MT-SDREM to the popular TF prediction tool oPossum 62 ., Our primary goal when comparing MT-SDREM with oPossum is to highlight the fact that using network information in the multi-task learning framework is useful ., The input to oPossum is a list of genes identified by the experiment ( s ) and using this list it attempts to find overrepresented TF-binding sites ., To select a common gene list from all three experiments we ranked the genes for each condition according to their differential expression and then merged the 3 rankings using the Kemeny-Young method 63 ., Similar to the number of genes used by MT-SDREM we used the top 3000 in the joint ranking as input to oPossum ., In Table 2 we present the comparison ., Note that since we used oPossum as the tool for the comparison of MT-SDREM with other methods for integrating data from several conditions , the results shown for Table 2 are different from the intersection results of Table 1 ., Here , for the MT-SDREM rankings we used the sum of % path flow going through each gene across the 3 networks to rank TFs ( Methods ) ., The oPossum TFs are ranked according to their Z-score ., While oPossum is able to identify a few relevant TFs , for most of the TFs identified by oPossum , we could not find significant roles in immune response regulation for them ., In contrast , several of the shared MT-SDREM TFs that are not identified by oPossum are known to play major roles in immune response as discussed above ., These include STAT1/3 , JUN/ATF2 , CEBPA/B which regulate a large number of immune response genes , RB1 which has been implicated in viral immune response networks 60 , PPARG , and SMAD ., MT-SDREM also uniquely identifies IRF1 which plays a major role in viral immune response by regulating interferons ., oPossum was able to identify only two relevant TFs that were not found by MT-SDREM ., These are ZEB1 which regulates the IL2 interleukin , part of the immune response system and AHR , part of the ANTR-AHR complex ., See also Tables S13–S15 in S1 Text for condition-specific comparisons using oPossum ., We also tried to compare MT-SDREM with the Inferelator method 6 but following email discussions with the authors of that method determined that such comparison is not feasible since Inferelator requires expression data for a large number of conditions while we only had time series response for three types of infections ., To compare the GO enrichment of shared genes/proteins we examined the top 500 genes in the combined MT-SDREM network ( ranked using the same sum of % of path flow going through genes across the 3 networks as we did for the oPossum comparison ) with the top 500 genes from the combined ranking of the differentially expressed ( DE ) genes from each condition ( combined using the Kemeny-Young method as explained before ) ., We used FuncAssociate 64 , 65 to compute standard GO enrichment for the genes ., We found 3 categories , only 2 of which were immune response related for which the p-value for DE genes was but which were not present in the MT-SDREM list or if present , their p-value was ., The categories are listed in Table 4 ., However , for the vice versa comparison , we found a large number of categories for which the MT-SDREM p-value was but which were either not enriched for in the DE genes list ( most common outcome ) or if present , their p-value was ., A subset of the immune response related categories are listed in Table 3 ., Note that we find significant enrichment for a very varied set of immune response processes including T cell activation , cytokine production , activation of immune response , etc . as well as categories related to viral genome expression and positive regulation of viral process ., The DE genes list is only enriched for negative regulation of viral process and viral genome replication ., The complete set of the categories is in S45 Table ., To further compare methods that are based on joint expression analysis to those that are based on joint network learning we looked at the GO enrichment for the top 50 TFs identified by MT-SDREM and oPossum ., The top 50 TFs for MT-SDREM are ranked using the joint ranking based on path flow for the 3 conditions as done for the GO comparison above ., We used the TF Z-score provided by oPossum to rank TFs for oPossum ., We again used FuncAssociate 64 , 65 to compute standard GO enrichment for the TFs ., We obtained only one immune-response related category ( interleukin related ) for which the p-value for the oPossum TF set was while being for MT-SDREM ( presented in Table 6 ) ., However we obtained 270 categories in total for which the MT-SDREM p-value was but the p-value for oPossum was , a large number of which were immune response related ., Due to space constraints , only a subset of these are presented in Table 5 ., These include postive regulation of innate immune response , viral process , and cytokine-mediated signaling pathway ., The complete list of categories is in S46 Table ., See also Tables S14–S25 in S1 Text for several additional comparisons of MT-SDREM and other methods using GO enrichment data ., Using the screen hit data for H1N1 and H5N1 we compared the performance of MT-SDREM , I-SDREM and Endeavour 66 , 67 ., Endeavour is a gene prioritization algorithm which uses a set of seed genes ( the sources ) to rank genes based on several types of evidence including gene expression , interaction networks derived from various sources , text mining , sequence similarity , and functional annotations ., It combines the individual rankings to create a global ranking for all genes ., For the MT-SDREM and I-SDREM results we ranked proteins based on the total number of paths weighted by their score going through them ., See Supplementary Methods in S1 Text for details ., For Endeavour , we configured it to use only BioGRID and HPRD as data sources as those are the only sources we use to construct our PPI network ., The expression data is not used by Endeavour ., We gave the source proteins as the seed genes to Endeavour ., We further compared these three methods with a baseline method that is condition-independent: ranking nodes by their weighted degree in the PPI network ., The results are presented in Figure 3 ., For H1N1 , the top 100 genes in the Endeavour ranking include only 20 screen hits ( p-value is 4 . 9E-7 ) ., For I-SDREM the number increases to 35 ( p-value 2 . 0E-19 ) whereas MT-SDREM obtains the highest number of protein in the overlap 39 ( p-value 1 . 7E-23 ) ., The baseline comparison where we rank by degree has an overlap of 30 genes ( p-value 9 . 4E-15 ) ., For H5N1 , the top 100 genes for Endeavour and for ranking by degree include only 5 screen hits ( p-value 1 . 2E-6 ) whereas both I-SDREM and MT-SDREM have an overlap of 9 screen hits ( p-value 1 . 7E-13 ) ., See also S1 Text for comparison of RNAi screen hits using GSEA ., We also compared MT-SDREM , I-SDREM with GeneMania 68 , 69 and concluded that MT-SDREM greatly improves upon the GeneMania results ., See Supplementary Results in S1 Text for details ., We developed MT-SDREM a multi-task learning framework that simultaneously reconstructs signaling and dynamic regulatory networks across related conditions ., Given the small number of condition-specific samples that are often available ( i . e . time series expression data and host-pathogen interaction data ) sharing parameters across related conditions allows the reconstruction of more accurate networks while still retaining the ability to explain condition-specific signaling and regulation ., We applied MT-SDREM to reconstruct networks for 3 related influenza A virus infections – H1N1 , H3N2 , and H5N1 ., The resulting signaling and regulatory networks were able to identify several known and novel regulators of immune and viral response ., Many of these were shared between all condition including PPARG , FOS , ATF , and JUN ., Similarly , we identify key signaling proteins , some shared by all conditions while others are unique to one or two of the conditions ., Specifically , we identified the signaling protein SUMO1 as part of pathway from UBE2I for all 3 conditions ., This agrees with recent findings that UBE2I interacts with SUMO1 to degrade influenza As virus , NS1 which is present in all three strains 70 ., We also identified the AKT1 gene , part of the PI3K/AKT pathway that is activated by NS1 in all conditions ., MT-SDREM is the first method to jointly reconstruct such dynamic networks ., Comparing MT-SDREM with methods that have been suggested to integrate gene expression data or with methods reconstruct such networks independently for each condition highlighted the advantages of multi-task network learning ., MT-SDREM outperformed previous methods in identifying a set of TFs controlling immune response , a set of functionally relevant proteins and a set of proteins whose knockdown affects viral loads ., While MT-SDREM can successfully utilize experiments from similar conditions to reconstruct signaling and regulatory networks , there are still issues we would like to improve in future work ., One direction we intend to explore is extending MT-SDREM to allow time based ( as opposed to global ) sharing of TFs across conditions so that splits representing the same time will be more likely to share TFs compared to other splits ., We would also like to improve on the models by using additional types of data , including epigenetic data which can help improve the priors for TF binding at specific time points by making them a function of the epigenetic code ., We use to denote the set of conditions that are jointly modeled by MT-SDREM ., Below we list the datasets used by MT-SDREM ., MT-SDREM extends the Signaling and Dynamic Regulatory Events Miner ( SDREM ) which has so far only been applied to reconstruct response networks for a single condition at a time 8 ., Prior to discussing the multi-task learning procedures we first briefly discuss the SDREM method ., SDREM is an iterative procedure that combines regulatory and signaling network reconstruction to model response pathways ., For the regulatory part , SDREM uses time series gene expression data with protein-DNA interaction data to identify bifurcation events in a time series ( places where the expression of previously co-expressed set of genes diverges – see Figure 2 ) , and the transcription factors ( TFs ) controlling these split events ., While some TFs are transcriptionally activated , others are only activated post-translationally via signaling networks ., To explain these TFs , the second part of SDREM links sources ( host proteins that directly interact with the virus/treatment ) to the TFs determined to regulate the regulatory network ., This part of SDREM uses protein-protein interaction ( PPI ) and protein modification data to infer such pathways – while imposing the constraint that the direction of PPI in the inferred pathways is consistent ., These two parts ( regulatory and signaling reconstruction ) iterate a fixed number of times until the final network is obtained ., See 8 for complete details ., Following the multi-task learning procedure we arrive at directed , weighted networks for each of the conditions being studied ., To further select the key proteins from each of these networks we rank the proteins based on the path flow going through a node ., The path flow through a node is defined as follows –where is the set of paths containing node ., To combine the rankings from each condition into a single ranking , we compute the total flow through all the nodeswhere is the set of genes and is the condition and then we computed the % flow through a node ., To get the combined score for a gene across conditions , we sum up the condition-specific % flows to get where is the number of conditions ., Then we rank the genes in descending order of the final score . | Introduction, Results and Discussion, Materials and Methods | Reconstructing regulatory and signaling response networks is one of the major goals of systems biology ., While several successful methods have been suggested for this task , some integrating large and diverse datasets , these methods have so far been applied to reconstruct a single response network at a time , even when studying and modeling related conditions ., To improve network reconstruction we developed MT-SDREM , a multi-task learning method which jointly models networks for several related conditions ., In MT-SDREM , parameters are jointly constrained across the networks while still allowing for condition-specific pathways and regulation ., We formulate the multi-task learning problem and discuss methods for optimizing the joint target function ., We applied MT-SDREM to reconstruct dynamic human response networks for three flu strains: H1N1 , H5N1 and H3N2 ., Our multi-task learning method was able to identify known and novel factors and genes , improving upon prior methods that model each condition independently ., The MT-SDREM networks were also better at identifying proteins whose removal affects viral load indicating that joint learning can still lead to accurate , condition-specific , networks ., Supporting website with MT-SDREM implementation: http://sb . cs . cmu . edu/mtsdrem | To understand why some flu strains are more virulent than others , researchers attempt to profile and model the molecular human response to these strains and identify similarities and differences between the resulting models ., So far , the modeling and analysis part has been done independently for each strain and the results contrasted in a post-processing step ., Here we present a new method , termed MT-SDREM , that simultaneously models the response to all strains allowing us to identify both , the core response elements that are shared among the strains , and factors that are uniquely activated or repressed by individual strains ., We applied this method to study the human response to three flu strains: H1N1 , H3N2 and H5N1 ., As we show , the method was able to correctly identify several common and known factors regulating immune response to such strains and also identified unique factors for each of the strains ., The models reconstructed by the simultaneous analysis method improved upon those generated by methods that model each strain response separately ., Our joint models can be used to identify strain specific treatments as well as treatments that are likely to be effective against all three strains . | genome expression analysis, medicine and health sciences, genetic networks, transcription activators, rna interference, genome scans, gene regulation, influenza, protein interaction networks, immunology, dna-binding proteins, microbiology, signaling networks, human genomics, orthomyxoviruses, dna transcription, viruses, infectious disease immunology, mathematical computing, h5n1, lymphocyte activation, rna viruses, scale-free networks, population modeling, genome analysis, transcription factors, network analysis, epigenetics, h1n1, directed graphs, gene ontology associations, gene ontology annotations, infectious diseases, influenza a virus, computer and information sciences, genomics, inflammation, proteins, medical microbiology, gene expression, microbial pathogens, regulatory networks, computing methods, proteomics, infectious disease modeling, graph theory, gene ontologies, immune response, rna interference screens, systems biology, biochemistry, avian influenza a viruses, computer modeling, genetics of the immune system, cell biology, clinical immunology, influenza viruses, gene regulatory networks, data visualization, viral pathogens, transcriptome analysis, genetics, biology and life sciences, viral diseases, computational biology, gene prediction, genomics statistics, organisms | null |
journal.pcbi.1001099 | 2,011 | Training Signaling Pathway Maps to Biochemical Data with Constrained Fuzzy Logic: Quantitative Analysis of Liver Cell Responses to Inflammatory Stimuli | Signaling networks regulate cell phenotypic responses to stimuli present in the extracellular environment 1 ., High throughput “interactome” data provide critical information on the composition of these networks 2 , 3 , 4 , but understanding their operation as signal processing systems is strongly advanced by direct interface with dedicated experimental data representing measured responses of biochemical species in the network ( proteins , mRNA , miRNA , etc . ) to stimulation by environmental cues in the presence or absence of perturbation 5 , 6 , 7 , 8 ., Immediate early responses are dominated by protein post-translational modifications ( we focus here on phosphorylation ) , assembly of multi-protein complexes , and changes in protein stability and localization ., Such responses are typically highly context dependent , varying with cell type and biological environment ., A critical question for the field is how large scale measurements of these responses can be combined with a signed , directed protein signaling network ( PSN ) to better understand the operation of complex biochemical systems 9 ., PSNs are typically deduced by manual or automated annotation of the literature ( e . g . 10 ) or directly from high-throughput experimental data ( e . g . 11 , 12 , 13 ) using a variety of computational techniques ., PSNs are represented as node-edge graphs 14 , and although they provide high-level insight into the composition and topology of regulatory networks 15 , 16 , 17 , 18 , 19 , 20 , as currently constituted PSNs are not readily ‘computable’ in that they cannot be used to calculate activation states of the key proteins in a pathway given a set of input cues , nor can quantitative relationships between pathways be determined ., This restricts the utility of PSNs for explicit prediction of responses and makes it difficult to compare network representations to functional experimental data ., A chief motivation of our current work is to determine how information encoded in a PSN can be made computable and compared to experimental data from a specific cell type , resulting in a context-specific network model ., Logic-based models ( e . g . 21 , 22 , 23 , 24 , 25 , 26; reviewed in 27 , 28 ) offer one means for converting interaction maps into computable models ., We have previously used Boolean logic ( BL ) to convert a literature-derived signed , directed PSN ( comprising for this purpose a ‘prior knowledge network’ PKN ) into a computable model that could be compared to experimental data consisting largely of the phospho-states of signal transduction proteins in the presence of different ligands and drugs 29 ., This approach allowed us to determine which links in the PKN were supported by the data , and generated models that were useful in making predictions about network topology 29 and drug targets 30 ., However , Boolean logic has a significant limitation , since real biochemical interactions rarely have simple on-off characteristics assumed by Boolean logic ., Thus , we require a means to encode graded responses and typical sigmoidal biological relationships in a logic-based framework ., One way to accomplish this is to apply traditional fuzzy logic FL , as demonstrated previously in modeling continuous input-output relationships to encode a complex signaling network 31 , 32 ., In the realm of control theory , FL modeling is an established technique for predicting the outputs of complex industrial processes when the influences of inputs cannot be characterized precisely 33 , 34 , 35 ., A central feature of FL is that it accounts for graded values of process states using a virtually unlimited repertoire of relationships between model species or components ., However , for past application to biochemical signaling networks , the flexibility of conventional FL modeling necessitated that the network topology be fixed prior to either manual 31 or computational 32 parameter fitting , rendering a formal training of network topology to experimental data infeasible ., In this paper we develop and employ a new approach to fuzzy logic modeling of biological networks that we term ‘constrained fuzzy logic’ cFL for descriptive purposes ., A key feature of cFL modeling is that it limits the repertoire of relationships between model species , enabling the formal training of a PKN to experimental data and resulting in a quantitative network model ., To maximize broad dissemination across the computational biology community , we implement cFL in an exisiting software tool CellNetOptimizer v2 . 0 ( CellNOpt ) , significantly extended to accommodate the further requirements of cFL while maintaining the BL analytic approach ( freely available at http://www . ebi . ac . uk/saezrodriguez/software . html ) ., We demonstrate the value of the CellNOpt-cFL method by elucidating new information from a recently published experimental dataset describing phospho-protein signaling in HepG2 cells exposed to a set of inflammatory cytokines 36 ., We show that a cFL model can be trained against a dataset and then validated by successful a priori prediction of test data absent from the training data ., We also establish the benefits of cFL relative to BL in three key areas:, ( a ) generation of new biological understanding;, ( b ) quantitative prediction of signaling nodes; and, ( c ) modeling quantitative relationships between signaling and cytokine release nodes ., Particular examples of validated biological predictions include:, ( i ) TGFα-induced partial activation of the JNK pathway and, ( ii ) IL6-induced partial activation of multiple unexpected downstream species via the MEK pathway ., Our work demonstrates the technical feasibility of cFL in modeling real biological data and generating new biological insights concerning the operation of canonical signaling networks in specific cellular contexts ., Fuzzy logic is a highly flexible methodology to transform linguistic observations into quantitative specification of how the output of a gate depends on the values of the inputs 33 , 37 , 38 , 39 ., For example , in the simplest , ‘Sugeno’ form of fuzzy logic , one specifies the following: ‘membership functions’ designating a variable number of discrete categories ( “low , medium , high , etc . ) as well as what quantitative value of a particular input belongs either wholly or partially to these categories; ‘rules’ designating the logical relationships between the gate inputs and outputs; AND and OR ‘methods’ designating the mathematical execution of each logical relationship; ‘weights’ designating the credence given any rule; and ‘defuzzification’ designating a scheme for determining a final output value from the evaluation of multiple rules 40 ., This flexibility is important in industrial process control 41 , which aims to use uncertain and subjective linguistic terms to predict how a controller should modulate a process variable to achieve the desired output ., However , our goal is to train models on quantitative biological data that are inevitably incomplete in the sense that, ( i ) measurements are not obtained under all possible conditions and, ( ii ) available data are not sufficient to constrain both the topology and quantitative parameters of the underlying networks ., Accordingly , we sought to develop a fuzzy logic system that minimizes the number of parameters to avoid over-fitting and simplifies the logic structure to facilitate model interpretability ., Because we aim to represent relationships among proteins in enzymatic cascades , mathematical relationships should be biologically relevant ., We therefore use a simple Sugeno fuzzy logic gate with a defined form ( see Text S1 ) based on transfer functions ( mathematical functions describing the relationship between input and output node values ) that approximate the Hill functions of classical enzymology ., Our ‘constrained’ fuzzy logic ( cFL ) framework uses a simplified fuzzy logic gate that is best described by the mathematical representation in Figure 1 ., The value of an output node of a one-input positive interaction is evaluated using a transfer function ., In this paper ‘input-output’ refers to the nodes of a specific cFL logic gate , where ‘nodes’ are molecular species ., We use the terms ‘model inputs’ and ‘model outputs’ to denote the overall relationship between model inputs such as ligand stimulation of cells and the collective output of the network ( protein modifications or phenotypic states in our application ) ., The transfer function underlying cFL gates is a normalized Hill function with two parameters: ( 1 ) the Hill coefficient , n , which determines the sharpness of the sigmoidal transition between high and low output node values and ( 2 ) the sensitivity parameter , k , which determines the midpoint of the function ( corresponding to the EC50 value in a dose-response curve , Figure 1a ) ., A negative interaction is represented similarly , except that the transfer function is subtracted from one , effectively inverting it ( Figure 1b ) ., Varying these parameters allows us to create a range of input-output transfer functions including linear , sigmoidal and step-like ( Figure 1a ) ., Moreover , this transfer function is biologically relevant: protein-protein interactions and enzymatic reactions can be described by Hill function formulations to a good approximation 42 , ., In some cases , use of a normalized function is too restrictive for practical application ., For example , if model inputs are purely binary ( values of either zero or one ) , the output of a normalized function would also be zero or one , making it impossible for a cFL gate to achieve intermediate states of activation ., Accordingly , our cFL method allows for alternative transfer functions ., For example , although the method is not limited to binary model inputs , the ligand inputs of our current work are binary ( either present or not ) ., If we used normalized transfer functions to relate these model inputs to downstream outputs , all model species would also be either zero or one ., Thus , for these transfer functions , we used a constant multiplied by the binary ligand input value ( see Materials & Methods ) ., If more than one input node influences an output node , this relationship is categorized as either an “AND” or “OR” interaction ., An AND gate is used when both input nodes must be active to activate the output node , whereas an OR gate is used when either input node must be active ., Mathematically , we represent AND behavior by evaluating each input-output transfer function and selecting the minimal possible output node value ( i . e . , applying the “min” operator , Figure 1c ) whereas we select the maximal value ( “max” operator; Figure 1d ) to evaluate an OR gate ., Finally , if both AND and OR gates are used to relate input nodes to an output node , our formalism evaluates all AND gates prior to OR gates ., This order of operations corresponds to the disjunctive normal or sum of products form 45 ., The process of training a cFL network ( CellNOpt-cFL ) has two starting requirements ., The first is a prior knowledge network ( ‘PKN’; Figure 2 , box A ) ., A PKN depicts interactions among the nodes as a signed , directed graph ( such as a PSN ) and can be obtained directly from the literature ., Alternatively , a large number of commercial ( e . g . , Ingenuity Systems: www . ingenuity . com; GeneGo: www . genego . com ) or academic ( e . g . , Pathway Commons: www . pathwaycommons . org , reviewed in 46 ) pathway databases as well as integrative tools ( e . g . 47 , 48 ) can be utilized to construct a PKN ., The second requirement is a dataset describing experimental measurements characterizing node activities following stimulation of and/or perturbations in upstream nodes ( ligand and inhibitor treatment in our example; Figure 2 , box B ) ., CellNOpt-cFL is then used to systematically and quantitatively compare the hypothesized PKN to the experimental dataset ., In practice , available experimental data is usually insufficient to fully constrain both the parameters and topology of the cFL models , and CellNOpt-cFL recovers many models that describe the data equally well ., Due to this typical absence of firm structural and parametric identifiability 29 , 49 , 50 , we examine families of models that fit the data equally well rather than attempting to identify a single global best fit ., Specifically , we examine interactions in the PKN that were either retained or consistently removed by training ., We also use individual models to predict input-output characteristics ., This treatment allows us to calculate both an average prediction as well as a standard deviation , which we show below can be useful for discrediting inaccurate predictions ., Our method comprises three main stages ( Figure 2 ) : first , structure processing converts a PKN into a cFL model; second , model training trains the model to experimental data; and third , model reduction and refinement simplifies trained models ., To illustrate CellNOpt-cFL , we examine a simple toy problem of training a PKN of the phospho-protein signaling network response to TGFα and TNFα ( Figure 2a . i ) to in silico data of activation of several downstream kinases in response to these ligands in the presence or absence of PI3K or MEK inhibition ( Figure 2a . ii ) ., In the first step , we streamline the network to contain only measured and perturbed nodes as well as any other nodes necessary to preserve logical consistency between those that were measured or perturbed ( 29; Figure 2 , Step 1 ) , resulting in a compressed PKN ( Figure 2 box C ) ., In our example , many nodes that were in the original PKN were neither measured nor perturbed experimentally ., Because these nodes could be removed without causing logical inconsistencies , they were not explicitly included in the compressed network ( Figure 2b ) ., In the second step , we expand the network into the multiple logical relationships ( combinations of AND and OR gates ) that can relate output nodes to their input nodes ( Figure 2 , Step 2 ) ., For example , our toy PKN was expanded to include all possible two-input AND gates governing the response of nodes with more than one possible input node ( Figure 2c ) ., In the third step , we train the cFL models to the data ( Figure 2 , Step 3 ) ., We start by limiting the possible parameter combinations to a subset of discrete parameter values that specify seven allowed transfer functions as well as the possibility that the input does not affect the output node ( i . e . the cFL gate is not present ) ., A discrete genetic algorithm determines transfer functions and a network topology that fit the data well by minimizing the mean squared error ( MSE , defined in Materials & Methods ) with respect to the experimental data ., Due to the stochastic nature of genetic algorithms , multiple optimization runs return models with slightly different topologies and transfer function parameters that result in a range of MSEs ., Models with an MSE significantly higher than the best models are simply eliminated from further consideration ., Models with similar MSEs but different topology and parameters result from the insufficiency of the data to constrain the model such that each model fits the data well albeit with slightly different features ., We consider each individual in this group as a viable model , and all are included for subsequent analysis ., Thus , after multiple independent optimization runs using the discrete genetic algorithm to train the expanded PKN against the data , a family of models with transfer functions chosen from a discrete number of possibilities is obtained ., For each of these models , we generate unprocessed models ( Figure 2 , box F ) by removing all cFL gates that are logically redundant with other cFL gates ( e . g . , in the gate “ ( B AND C ) OR B activate D” , the AND gate is logically redundant with the “B activates D” gate ) ., These gates are removed because they increase model complexity by using multiple logic gates to encode a relationship that can be specified by a simpler gate ., In our toy example , a family of twenty unprocessed models was obtained by training the expanded map ( Figure 2c ) to in silico data ( Figure 2a . ii . ) using the discrete genetic algorithm ., The unprocessed models from different optimization runs had similar topologies with the exception of the gate describing the relationship of MEK to its input nodes: TGFα and Akt ( Figure 2d , brown and green dashed gates ) ., Sixteen of the unprocessed models described the activation of MEK as depending only on TGFα ( brown , dashed gate ) whereas four described activation using the AND NOT gate ( green , dashed gate ) ., In the model reduction and refinement stage ( Steps 4–6 ) , we determine which gates can be removed altogether as well as AND gates that can be replaced with one-input cFL gates without significantly affecting the MSE ., We implemented the non-exhaustive heuristic search procedure described below on each unprocessed model and illustrate its application to our toy example ( Figure 3 ) ., In the fourth step , we remove or replace all gates for which the alteration does not increase the MSE of the unprocessed model over some threshold , which we term the ‘reduction threshold’ ., We use a range of reduction thresholds such that each unprocessed model results in several models , one for each reduction threshold used ., Following this step , the resultant models are considered reduced models ., In the fifth step , we fix the model topology to that obtained during Step 4 and treat the transfer function parameters in each reduced model ( Figure 2 , Step 5 ) as continuous parameters rather than the discrete set of transfer function parameters required for use of the discrete genetic algorithm ., We use a Sequential Quadratic Programming method ( Text S1 ) to refine the model parameters and further improve the fit of the models to the experimental data ., The resulting models are termed reduced-refined models , which have a range of MSEs depending on the reduction threshold used ( Figure 3a ) ., In the sixth and final step , we specify a reduced-refined model to represent each unprocessed model ( Figure 2 , Step 6 ) ., For each unprocessed model , we choose the reduced-refined model that has the fewest number of fitted transfer function parameters without increasing the MSE above a defined ‘selection threshold . ’ The selection threshold is chosen by comparing the average number of parameters in the family of models to the average MSE of the models ( Figure 3b ) ., The net result is a set of reduced-refined-filtered models ( hereafter referred to as filtered models , Figure 2 , Box G ) ., In our toy example , the filtered models have identical topology and in no case does Akt inhibit MEK activation ( Figure 2e ) ., This topology is , in fact , the topology from which the in silico data was derived ., The ability of cFL to fit intermediate values made it possible to recover the correct model topology , whereas BL did not identify the correct model , and a gate linking TGFα to PI3K was consistently missing ( Figure 2e , dashed arrow ) ., Specifically , BL was unable to return the correct topology because nodes downstream of PI3K ( Akt and JNK ) were partially activated ( 0 . 32 and 0 . 19 , respectively ) under conditions of TGFα stimulation , and a BL model that included the TGFα to PI3K gate had a higher error ( MSE =\u200a0 . 56 ) than a model that omitted the interaction ( MSE =\u200a0 . 07 ) ., In contrast , the improved ability of cFL to model graded activities made it possible to recover the true network topology ., While the expansion step ( Figure 2 , step, 2 ) captures the many possible combinations of AND and OR logic relationships between nodes , it also increases the complexity of the network , resulting in an increase in the size of the optimization problem ., Depending on the biological network of interest , some or most of these AND gates might not be biologically relevant ., For example , it is unlikely that six receptors must be active in order to activate another species , as would be the case for a six-input AND gate ( instead , it is more likely to be a OR gate ) ., A profusion of AND gates also makes the resultant networks difficult to interpret because most AND gates are in only a few models whereas the majority of models contain single-input and OR gates ., Thus , the AND gates can effectively appear as system “noise” , interfering with visual assessment as well as computational analysis of the model topologies ., Because of these potential complications , the expansion step can be limited to include only AND gates with a few inputs , depending on the complexity one would like to capture with the trained network models ., In the current paper , we have limited the search in the discrete genetic algorithm to a set of seven transfer functions ., Use of more or fewer transfer functions is possible , but we found that seven transfer functions allowed us to represent a variety of input-output relationships without unduly increasing problem complexity to the point that the discrete genetic algorithm no longer consistently returned models that fit the data well ( see Materials & Methods ) ., To test the ability of cFL modeling to analyze real biological data , we modeled a set of measurements describing the response of the HepG2 hepatocellular carcinoma cell line to various pro-survival , pro-death , or inflammatory cytokines in the presence or absence of specific small molecule kinase inhibitors ., This dataset was used to construct a recent BL model 29 ., Here we ran an independent analysis using the cFL approach and compare the results to the BL previously reported ., The dataset comprises measurement of phosphorylation states as markers of activation of 15 intracellular proteins before and 30 minutes after stimulation by one of six cytokines in the presence or absence of seven specific small molecule kinase inhibitors ( Figure 4a , Figure S1 ) ., The measurements were normalized to continuous values between zero and one using a routine implemented in the MATLAB toolbox DataRail 51 , as previously described ( 29 , see Text S1 ) ., The HepG2 dataset was trained to several related PKNs which are enumerated in Table 1 and Figure S2 ., These PKNs were derived , with various extensions , from the Ingenuity Systems database ( www . ingenuity . com ) with manual addition of literature data about IRS1 that was obviously missing 29 ., The first PKN , termed PKN0 was identical the one used previously for BL modeling 29 ., In the course of our analysis , we found it necessary to search the literature for interactions missing in PKN0 but supported by the data , resulting in several PKNs ( Table 1 ) ., Furthermore , we limited the manner in which the PKNs were expanded in two ways: ( 1 ) expansion into all possible two-input AND gates or ( 2 ) expansion into a two-input AND gate only when one input was inhibitory ., In the second case , the expansion of inhibitory gates was necessary because , in logic terms , an inhibitory gate indicates that the output node is active when the input node is not active ., In biological networks , this is true if the output node is constitutively active , which was not observed in the normalized HepG2 data ., Thus , in order to accurately model the inhibitory effect , it had to occur in conjunction with activation by some other input node , which is captured by an AND gate ., If a PKN was processed with both types of expansion , we include a superscript to differentiate between the two cases – i . e . , PKN1a for the expansion of all gates and PKN1i for the expansion of only the inhibitory case ., PKN0 was expanded to include all possible two-input AND gates and trained to the HepG2 dataset with CellNOpt-cFL ( Figure S2 ) ., The 90 unprocessed cFL models obtained after training showed that PKN0 exhibited a poor fit to IL1α-induced protein phosphorylation ( Figure S3 ) , a result we had also observed with BL analysis 29 , confirming that the poor fit of BL was due to errors in the topology of PKN0 and not the inability of Boolean logic to fit intermediate values ., An inspection of systematic model/data disparity ( Figure S3 ) immediately indicated that the models did not fit IL1α-induced phosphorylation of IRS1 , MEK and several species known to be modulated by the MEK pathway ., In PKN0 , no paths between IL1α and MEK or IRS1 were present ., Based on careful reading of the literature , we added two links to PKN0: a TRAF6 → MEK link 52 , and an ERK → IRS1 link 53 ., These links had been inferred by the BL framework 29 and were supported by further literature evidence ., To add a link that provided a path between IL1α and MEK in the absence of BL inference results , for simplicity one should first consider links from species that IL1α is already known to activate ., In this case , TRAF6 is the most upstream species which experimental evidence suggests can activate MEK 52 ., In the case of IRS1 signal activation , the specific phosphorylation site measured should be considered ., Our data included measurements of phospho-S636/639 , and S636 is a known phosphorylation site of ERK2 53 ., A novel finding from CellNOpt-cFL analysis of the HepG2 data was that IL6 treatment led to phosphorylation of several downstream proteins ., Similarly to the links just considered , PKN0 included no paths between IL6 stimulation and these downstream proteins , resulting in an inability to fit this pattern of phosphorylation ., Importantly , however , BL analysis would not have recognized this partial activation due to its inability to fit intermediate values ( as illustrated in our earlier toy example ) ., Because IL6 was observed to partially activate Akt in the data and known mechanisms exist for this activation 54 , we added a prospective IL6R → PI3K link to the PKN , thus providing an extended PKN ( PKN1 ) that we use below for subsequent CellNOpt-cFL analysis ., PKN1 was expanded to include all possible two-input AND gates ( PKN1a ) for a total of 170 discrete parameters corresponding to 105 logic gates ., The resultant network was trained to the HepG2 data ., Reduction of the PKN1a–derived models indicated that almost all AND gates could be removed or replaced by single-input gates ., Since the AND gates appeared to add unnecessary complexity to the cFL models , we also expanded PKN1 to only include AND gates if an input node was inhibitory ( PKN1i; Table 1 ) , resulting in only 60 discrete parameters corresponding to 56 logic gates ., We then compared the PKN1a- and PKN1i-derived cFL models ., The comparison of these two PKN-derived model families revealed a clear tradeoff between model fit and complexity ., The more complex PKN1a-derived models were able to fit the data slightly better than the PKN1i-derived models ( average unprocessed model MSE of 0 . 032±0 . 002 compared to 0 . 035±0 . 002 , p<0 . 001 ) ., However , the more complex PKN1a-derived models contained many more parameters than the PKN1i-derived models both before and after optimization ( 170 compared to 60 discrete parameters before optimization and an average of 72 . 8±4 . 9 compared to 66 . 6±3 . 9 continuous parameters after optimization ( p<0 . 001 ) ; Figure S4 ) ., The simpler PKN1i-derived models used fewer initial and final parameters to arrive at a fit to the data only 9% worse than PKN1a-derived models ., Since the 9% deviation is in the range of error in the normalized data ( error estimated to be 10% by comparing similar stimulation conditions ) , we focused subsequent analysis on the simpler PKN1i-derived models ., For completeness , we include the results of PKN1a-derived models as supplemental information ( Figure S5 ) ., To determine the statistical significance of our results , we compared the family of 243 unprocessed models with unprocessed models obtained from either training PKN1i to randomized data or training a randomized PKN1i to the data ( Table S1 ) ., Data was randomized by pairwise exchange of all data values while network topologies were randomized either by generation of an entirely random topology or by random pairwise exchange of gate inputs , gate outputs , or nodes inputs 29 ., When compared to the results of all types of randomization , models trained to the real data and PKN1 were highly significant ( P-value <0 . 001 , Table S1 ) , indicating that the family of trained cFL models fit the data better than expected by random chance ., To probe the dependence of the CellNOpt-cFL training process on the quality of the PKN used , we randomly added links to or removed links from the PKN and trained the resultant PKN to the data ., As expected , the models derived from PKNs with links randomly removed had a poorer fit to data than those derived from the complete PKN1i ( Figure 4b , solid line ) ., Conversely , when links were randomly added to the PKN , cFL-CellNOpt effectively removed the links ( Figure S6 ) , resulting in models with similar goodness of fit as models derived from PKN1i ( Figure 4b , dashed line ) ., We thus conclude that an incomplete PKN degrades the ability of CellNOpt-cFL to fit the data whereas models derived from a PKN with extraneous links retain this ability ., As an initial investigation of model predictive capacity and a check for over-fitting , we performed a ten-fold cross-validation by randomly dividing the HepG2 data into ten subsets and , for each subset , reserving one as a test set while training with the remaining nine data subsets ., The similar fits of the training and test data provided evidence that the family of models obtained from this procedure were predictive , and the difference in test and training MSEs did not depend on selection threshold , a measure of model size , suggesting that the models were not over-fit ( Figure 4c ) ., Analysis of this cross-validation result combined with a plot of average filtered model size and fit ( MSE ) as a function of selection threshold ( Figure 4d ) suggested that a selection threshold in the range 1×10−3 – 1×10−2 would result in a family of models that contain slightly fewer number of parameters than lower thresholds ( Figure 4d , dashed line ) while retaining the ability to fit the data well ( Figure 4d , solid line ) ., We used a threshold of 5 . 0×10−3 for the remainder of our analysis unless otherwise noted ., Finally , we obtain a family of 243 filtered models for further analysis ( Figure 5 ) ., By taking note of which cFL gates are removed during the CellNOpt-cFL training and reduction processes , one can generate hypotheses regarding these gates ., Table 2 summarizes a set of biological hypotheses readily suggested by our cFL model topologies ., Analysis of error between the family of cFL models and experimental data ( Figure S7 ) highlighted consistent error in TGFα-induced partial activation of c-Jun ., Both PKN0 and PKN1 allowed for TGFα-induced activation of c-Jun by the JNK pathway via crosstalk from Ras or PI3K to MAP3K1 ., In the BL methodology , this crosstalk was removed due to the inability to fit partial activation , and no BL model allowed for activation of c-Jun after TGFα stimulation ., However , we found that a subset of cFL models accounted for this c-Jun partial activation by including crosstalk between Ras or PI3K and MAP3K1 ., These models also partially activated JNK after TGFα stimulation , a feature that was inconsistent with the training data ( Figure S8 ) ., Thus , these models predict that JNK was actually phosphorylated under conditions of TGFα stimulation , but our measurements did not detect it ., To test this prediction directly , we undertook de novo measurement of JNK and c-Jun phosphorylation following stimulation with different doses of TGFα ( Figure 6a ) ., These new data show that JNK does indeed become phosphorylated upon stimulation of HepG2 cells with TGFα ., Thus , the cFL models containing crosstalk from Ras or PI3K to MAP3K1 were the correct models ., Combined with Table 2 , this analysis highli | Introduction, Results, Discussion, Materials and Methods | Predictive understanding of cell signaling network operation based on general prior knowledge but consistent with empirical data in a specific environmental context is a current challenge in computational biology ., Recent work has demonstrated that Boolean logic can be used to create context-specific network models by training proteomic pathway maps to dedicated biochemical data; however , the Boolean formalism is restricted to characterizing protein species as either fully active or inactive ., To advance beyond this limitation , we propose a novel form of fuzzy logic sufficiently flexible to model quantitative data but also sufficiently simple to efficiently construct models by training pathway maps on dedicated experimental measurements ., Our new approach , termed constrained fuzzy logic ( cFL ) , converts a prior knowledge network ( obtained from literature or interactome databases ) into a computable model that describes graded values of protein activation across multiple pathways ., We train a cFL-converted network to experimental data describing hepatocytic protein activation by inflammatory cytokines and demonstrate the application of the resultant trained models for three important purposes:, ( a ) generating experimentally testable biological hypotheses concerning pathway crosstalk ,, ( b ) establishing capability for quantitative prediction of protein activity , and, ( c ) prediction and understanding of the cytokine release phenotypic response ., Our methodology systematically and quantitatively trains a protein pathway map summarizing curated literature to context-specific biochemical data ., This process generates a computable model yielding successful prediction of new test data and offering biological insight into complex datasets that are difficult to fully analyze by intuition alone . | Over the past few years , many methods have been developed to construct large-scale networks from the literature or databases of genetic and physical interactions ., With the advent of high-throughput biochemical methods , it is also possible to measure the states and activities of many proteins in these biochemical networks under different conditions of cellular stimulation and perturbation ., Here we use constrained fuzzy logic to systematically compare interaction networks to experimental data ., This systematic comparison elucidates interactions that were theoretically possible but not actually operating in the biological system of interest , as well as data that was not described by interactions in the prior knowledge network , pointing to a need to increase our knowledge in specific parts of the network ., Furthermore , the result of this comparison is a trained , quantitative model that can be used to make a priori quantitative predictions about how the cellular protein network will respond in conditions not initially tested . | computational biology/systems biology, computational biology/signaling networks | null |
journal.pgen.1003518 | 2,013 | Histone Acetyl Transferase 1 Is Essential for Mammalian Development, Genome Stability, and the Processing of Newly Synthesized Histones H3 and H4 | The packaging of genomic DNA during replication is a highly orchestrated process that ensures both the necessary compaction of the DNA and the proper transmission of the epigenetic landscape 1 , 2 , 3 , 4 , 5 ., An important aspect of chromatin assembly is the processing of newly synthesized histones for their incorporation into chromatin ., The transient acetylation of histone H3 and H4 NH2-terminal tails is a hallmark of this processing ., Newly synthesized molecules of histone H4 are predominantly diacetylated ., This diacetylation occurs specifically on lysine residues 5 and 12 and this precise pattern is widely conserved throughout eukaryotic evolution ., The acetylation of histone H3 occurs on a smaller fraction of the newly synthesized molecules and does not occur in a consistent pattern across eukaryotes ., A role for this acetylation in histone deposition was first suggested by the correlation between the presence of these histone marks and active chromatin assembly as H3 and H4 are rapidly modified after their synthesis and then deacetylated following their incorporation into chromatin 6 ., However , despite this longstanding correlation , an understanding of the function of histone NH2-terminal tail domain acetylation in chromatin assembly remains elusive ., In addition to their NH2-terminal tail domains , evidence from S . cerevisiae indicates that newly synthesized histones are also acetylated in their core domains with H3 acetylated on lysine 56 and H4 acetylated on lysine 91 7 , 8 , 9 , 10 ., H3 lysine 56 lies near the entry/exit point of the nucleosome in close proximity to the DNA ., The acetylation of this site occurs specifically in S phase and has been linked to chromatin assembly by a number of observations ., First , mutations in yeast that alter H3 lysine 56 cause defects in the reassembly of chromatin structure that accompanies the recombinational repair of a DNA double strand break ., Second , H3 lysine 56 mutations influence the binding of histone H3 to the CAF-1 histone chaperone complex that plays a key role in replication coupled chromatin assembly 7 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ., Histone H4 lysine 91 lies in the interface between H3/H4 tetramers and H2A/H2B dimers where it forms a salt bridge with an aspartic acid residue in histone H2B ., Hence , the acetylation of H4 lysine 91 may regulate tetramer-dimer interactions and genetic results are consistent with a role for this modification in chromatin assembly 10 , 18 ., Enzymes known as type B histone acetyltransferases ( HATs ) catalyze the acetylation of newly synthesized histones ., Type B HATs are primarily distinguished from type A HATs by their substrate specificity ., As expected for enzymes that modify histones prior to their assembly into chromatin , type B HATs are highly specific for free histones ., Type B HATs may also function outside of the nucleus 19 ., A number of type B HATs have now been identified ., The first was Hat1p , which acetylates free histone H4 on lysine residues 5 and 12 20 , 21 ., In addition , the yeast enzyme Rtt109p acetylates free histone H3 on lysine 56 and lysine 9 22 , 23 , 24 ., Interestingly , the original type A HAT , Gcn5p , may also possess type B HAT activity in S . cerevisiae as it has been shown to be involved in the acetylation of the NH2-terminal tail of newly synthesized histone H3 25 , 26 ., Finally , the mammalian enzyme HAT4 may also be a type B HAT as it resides in the Golgi and is capable of acetylating histone H4 lysine 91 27 ., Originally isolated from budding yeast , Hat1p was found to exist in at least 2 complexes ., The first is a cytoplasmic complex that also contains Hat2p , which is a homolog of the Rbap46 histone chaperone in mammalian cells 21 , 28 ., Hat1p is also found in a nuclear complex that , in addition to Hat2p , contains another histone chaperone , Hif1p ( a homolog of the mammalian protein NASP ) 29 ., Not only the Hat1p protein itself but also the Hat1p-containing complexes are highly conserved in eukaryotes ., Complexes with similar compositions have been isolated from human , frog , chicken and corn ., As expected from their high degree of similarity , these enzymes specifically acetylate free histone H4 on lysines 5 and 12 30 , 31 , 32 , 33 ., Despite its widespread conservation , initial genetic analyses in yeast showed that loss of Hat1p had no detrimental effects on either chromatin assembly or cell viability 20 , 21 , 34 ., This lack of phenotypic effect was , at least partly , due to functional redundancy as combining the deletion of HAT1 with mutations in specific sets of lysine residues in the histone H3 NH2-terminal tail resulted in defects in telomeric silencing and DNA damage sensitivity 35 , 36 ., DNA damage sensitivity has also been observed in S . pombe and chicken DT40 cells lacking Hat1 37 , 38 ., Importantly , direct evidence linking yeast Hat1p to chromatin assembly in the contexts of DNA damage repair and histone exchange has recently been reported 39 , 40 ., Despite the minor effects on cell viability observed in the absence of Hat1 , biochemical analyses have implied that Hat1 may play a critical role in histone processing and dynamics ., This is suggested by an intriguing property of Hat1p ., Unlike most enzymes , Hat1 appears to remain stably associated with its histone substrates following acetylation 41 ., This property of Hat1p also appears to be widely conserved ., In yeast , both the cytoplasmic and nuclear Hat1p-containing complexes are stably associated with histones H3 and H4 29 ., In mammalian cells , Hat1p appears to be one of the primary proteins physically associated with soluble histones 42 , 43 , 44 , 45 , 46 , 47 ., Therefore , Hat1p has the potential to function both catalytically and stoichiometrically in the chromatin assembly process ., To explore the function of Hat1 and the acetylation of newly synthesized histones in mammals , we have generated a conditional Hat1 knockout mouse model ., Hat1−/− animals are neonatal lethal with developmental lung defects ., These result from hyper proliferation of mesenchymal cells leading to severe atelectasis , less aeration and death upon respiratory failure ., In addition , a significant fraction of the Hat1−/− animals display severe craniofacial defects ., Mouse embryonic fibroblasts ( MEFs ) derived from Hat1−/− embryos show multiple defects including slow growth , DNA damage sensitivity and genome instability ., Analysis of proteins present on newly replicated DNA by iPond ( isolation of proteins on nascent DNA ) indicates that histones H3 and H4 deposited during replication-coupled chromatin assembly are hypo-acetylated in the absence of Hat1 48 ., Consistent with these observations , analysis of newly synthesized histones indicates that Hat1 is the sole HAT responsible for the acetylation of newly synthesized histone H4 ., Surprisingly , loss of Hat1 also leads to a decrease in the modification of newly synthesized histone H3 ., These results demonstrate that Hat1 is essential in mammals and that it plays an integral role in the processing of newly synthesized histones during the process of chromatin assembly ., There is a single homolog of Hat1 in the murine genome that is highly similar to human and yeast Hat1 ., The murine Hat1 gene consists of 11 exons ( Figure 1A ) ., A construct was generated to target the integration of loxP sites to flank Hat1 exon 3 ., In the presence of cre recombinase , exon 3 can be deleted with the subsequent introduction of a stop codon ( Figure 1A ) ., This will create a truncation mutant of Hat1 that produces a product less 50 amino acids long ., In the event that alternate splicing occurs in the Hat1 gene that could skip exon 3 , only exon 9 can be spliced to exon 2 and retain the proper reading frame ., In this case , the protein that would be produced would lack the entire Hat1 active site ., The targeting construct was transfected into mouse embryonic stem ( ES ) cells and cells grown with antibiotic selection ., Cell lines in which the targeting construct was properly integrated were identified ( data not shown ) ., These cells were then injected into blastocysts to generate chimeric mice ., The chimeras were then mated with wild type mice ( C57/Bl6 ) and the pups were screened by Southern blot to determine whether germline transmission of the Hat1flox allele had been achieved ., Several animals with a Hat1flox/Hat1+ genotype were identified ( Figure 1B ) ., To generate a complete Hat1 mouse knockout , the Hat1flox/Hat1+ mice were mated to mice that ubiquitously express the Cre recombinase ., The litters from these matings were screened to identify offspring in which Hat1 exon 3 had been deleted ( Hat1Δ3 ) ., As seen in Figure 1C , several mice were obtained with the genotype Hat1Δ3/Hat1+ ., These mice also carried a copy of the Cre recombinase gene ., We backcrossed these mice with C57/Bl6 mice to obtain animals that no longer expressed Cre to avoid any undesired phenotypic consequences that could arise from expression of this recombinase ., Following backcrossing to remove cre from the genome , Hat1+/Hat1Δ3 mice were mated and the genotypes of the resulting pups were determined ( Figure 1D , E ) ., For simplicity , Hat1Δ3 mice will be referred to as Hat1− ., As seen in Figure 1D , based on the number of Hat1+/+ pups born , there were slightly fewer than expected Hat1+/− pups and a marked decrease in the number of Hat1−/− pups born ., Importantly , all of the Hat1−/− pups were either born dead or died shortly after birth ( Figure 1F ) ., In addition , the Hat1−/− pups were approximately 20% smaller than their Hat1+/+ counterparts ( Figure 1F ) ., Contrary to what is observed in the other model organisms that have been examined , Hat1 is necessary for viability in mice as the loss of this enzyme results in neonatal lethality ., To determine the cause of this lethality , Hat1+/+ and Hat1−/− neonates were subjected to pathological examination ., Significantly more cells per alveolar septum , which is a measure of fetal lung immaturity , were observed in the Hat1−/− neonates compared to the WT mouse lung ( Figure 2A ) ., The lungs from the neonatal Hat1−/− pups also showed a lower overall lung maturation , which was compiled by an assessment of vascularity , aerated lung tissue and septum thickness ., These defects in lung development resulted in atelectasis , less aeration and finally lead to respiratory failure ( Figure 2A ) ., Lungs of Hat1+/+ controls were completely normal ( Figure 2A ) ., Hat1 is highly expressed in alveolar as well as lung interstitial cells of Hat1+/+ mice ( Figure 2B ) ., A highly significant increase in Ki67+ proliferation rates was observed in Hat1−/− compared to Hat1+/+ neonates ., However cleaved Caspases 3 expression was not altered suggesting that the developmental defect was the result of inappropriate proliferation rather than a defect in apoptosis ( Figure 2C , D ) ., The inappropriate proliferation begins early in development and is apparent by 11 . 5 d . p . c . ( Figure S1 ) ., There was another more obvious , but less penetrant , phenotype observed in the Hat1−/− mice ., Approximately 25% of the Hat1−/− neonates were born with craniofacial abnormalities ., The skeletal structures of Hat1+/+ and Hat1−/− neonates were examined by microCT scanning ., As seen in Figure 3A , the structure of the skulls from a number of the neonates is highly abnormal ., For example , the nasal passages were often missing , having been overgrown with bone ., In addition , there were severe defects in the development of the lower jaw structure ., These defects included several neonates where the lower jaw was missing in its entirety ( Figure 3A , right-hand panels ) or examples where the lower jaw bones were fused into a single bone ( Figure 3A , middle panel ) ., Defects to the remainder of the skeletal system were less severe ., As shown in Figure 3B , while the upper areas of the spine are relatively normal , the structure of the vertebrae in the Hat1−/− neonates degenerate near the base of the spinal column ., To characterize the skeletal defects in more detail , WT and Hat1−/− neonates were stained with Alcian blue and Alizarin red ., Alcian blue stains cartilage blue and Alizarin red stains bone purple ., Consistent with the microCT results , bone and cartilage staining is similar for the majority of the skeletal system in the WT and Hat1−/− neonates ( Figure 3C ) ., However , there are differences in the head ., The Hat1−/− neonates showed a marked increase in bone density in the skull and a decrease in the amount of cartilage staining ., These results are consistent with the defects seen in lung development as the increased bone density may be the result of osteoblast hyper-proliferation in the absence of Hat1 ., To correlate embryonic expression of Hat1 with the skeletal defects observed in the Hat1−/− neonates , whole embryos were stained with α-Hat1 antibodies ., As seen in Figure 3D , there is widespread expression of Hat1 protein in the head of embryos ., There is also a high level of staining in the abdominal region ., Closer examination of Hat1 in the head by immunohistochemistry showed that Hat1 is widely expressed in most tissue types ( Figure 3E ) ., Therefore , the phenotypes observed in the Hat1−/− neonates are not strictly linked sites of Hat1 protein expression ., The fact that Hat1−/− offspring survive to at least late embryogenesis facilitated the generation of Hat1−/− embryonic fibroblast cell lines to address specific questions about the function of Hat1 in mammalian cells ., Mouse embryonic fibroblasts ( MEFs ) were generated from Hat1+/+ , Hat1+/− and Hat1−/− embryos ( Figure 4A ) ., Western blot analysis using α-Hat1 antibodies confirmed that the MEFs isolated from the Hat1−/− embryos were completely devoid of Hat1 protein ( Figure 4A ) ., In addition , heterozygous MEFs ( isolated from Hat1+/Hat1− embryos ) showed an ∼2-fold decrease in Hat1 protein levels ., The loss of Hat1 protein does not influence cell proliferation in any of the model organisms in which it has been genetically deleted ( yeast and avian cells ) ., To determine whether Hat1 was important for mammalian cell proliferation , growth curves were measured for the WT , heterozygous and Hat1 null MEFs ., As seen in Figure 4B , only minor differences in proliferation between the WT and heterozygous cells were observed ., However , Hat1 null cells showed an ∼50% decrease in cell proliferation ., To determine whether the decrease in cell proliferation seen with the Hat1−/− MEFs was the result of a specific defect in cell cycle progression , FACS analysis was used to monitor cell cycle distribution ., As seen in Figure 4C , Hat1−/− MEFs displayed a moderate accumulation of cells in G2/M suggesting that the decrease in cell proliferation seen in the absence of Hat1 may be , at least in part , due to a G2/M delay in these cells ., Taken together , these results indicate that Hat1 protein is not essential for the proliferation of mammalian cells but that cell cycle progression is defective in the absence of this enzyme ., Loss of Hat1 in budding yeast , fission yeast and chicken DT40 cells results in the sensitivity to DNA damaging agents 36 , 37 , 38 ., To determine whether a role for Hat1 in DNA damage repair is conserved in mammalian cells , WT and Hat1−/− MEFs were assayed for their sensitivity to a variety of DNA damaging agents ., To avoid complications arising from the limited proliferation potential of primary MEFs , immortalized cells lines from Hat1+/+ and Hat1−/− primary MEFs were generated via transfection with SV40 large T antigen ( proliferation rates of immortalized MEFs are shown in Figure S2 ) ., Equal numbers of Hat1 WT and Hat1 null cells were plated and allowed to grow in normal serum containing either methyl methane sulfonate ( MMS ) or hydroxyurea ( HU ) or following exposure to ultraviolet radiation ( UV ) ( Figure 4D ) ., The Hat1−/− cells showed a pronounced sensitivity to each of these DNA damaging agents ., Hence , Hat1 plays a critical role in DNA damage repair in mammalian cells ., Interestingly , the Hat1−/− mammalian cells were sensitive to a broader range of DNA damaging agents ., Both yeast and avian cells lacking Hat1 are sensitive to MMS but not UV radiation , suggesting that these Hat1−/− mutants are specifically sensitive to DNA double strand breaks 36 , 37 , 38 ., However , the Hat1−/− MEF cell lines sensitivity to both types of DNA damage indicating that Hat1 is important for multiple pathways of DNA repair ., As loss of Hat1 resulted in sensitivity to DNA damage , we next explored whether Hat1 was also necessary for proper genome stability ., One hallmark of genome instability is the presence of DNA damage in the absence of treatment with DNA damaging agents ., Hat1+/+ cells showed undetectable levels of endogenous DNA damage , as measured by the presence of γ-H2AX foci ( Figure 5A ) ., In contrast , untreated Hat1−/− cells showed numerous γ-H2AX foci ., An increase in γ-H2AX levels , both before and after DNA damage , in Hat1−/− MEFs was confirmed by Western blot analysis of whole cell extracts ( Figure S3 ) ., To directly visualize genome structure , we generated metaphase spreads from Hat1 WT and Hat1 null cells ., Genomic abnormalities that were more frequently observed in the Hat1−/− cells than in the Hat1+/+ cells were of several different types ( Figure 5B ) ., First , there was a significant increase in chromatid breaks and chromosome fusions ., Representative examples are shown in Figure 5C ( additional metaphase spreads are shown in Supplemental Figure S4 ) ., In this figure , black arrows indicate examples of chromatid breaks or gaps ., Red arrows indicate examples of chromosomal fusions ., These included both “end-to-end” fusions and unusual “bridge-like” structures where the ends of one chromosome were fused to internal regions of another chromosome ., We also observed changes in chromosome number ., There was an increase in the number of aneuploid cells that contained a smaller than normal number of chromosomes in the Hat1−/− cells ., In addition , there were high numbers of Hat1−/− cells with a 4n DNA content ( Figure 5C ) ., In summary , the absence of Hat1 resulted in the presence of high levels of endogenous DNA damage and chromosomal abnormalities , indicating that Hat1 plays an essential role in maintaining genome stability ., Since its initial discovery and biochemical characterization , Hat1 has been presumed to be involved replication-coupled chromatin assembly through the conserved diacetylation of newly synthesized histone H4 ., However , evidence to support this hypothesis has been circumstantial 41 , 49 ., The availability of mammalian cells genetically deleted for Hat1 allowed us to definitively address this issue ., To directly determine whether Hat1 is involved in the acetylation of histones that are incorporated during replication coupled chromatin assembly , we used iPond to monitor histone modification dynamics on newly replicated DNA 48 ., The iPond technique involves pulse-labeling cells with the thymidine analog EdU ., The EdU will then be incorporated into DNA that is synthesized during the pulse phase ., Following cross-linking , Click chemistry can then be used to covalently attach biotin to the EdU moieties , which allows for the affinity purification of the nascent DNA using streptavidin beads ., Western blot analysis of the fractions that elute from the streptavidin beads can then be used to monitor the presence of specific proteins or their modified isoforms on the newly synthesized DNA ., Immortalized Hat1+/+ and Hat1−/− MEFs were pulsed with EdU for 15 minutes and then chased with thymidine for 90 minutes ., The 90 minute thymidine chase allowed us to distinguish between stably associated chromatin proteins and proteins that associate with newly replicated DNA but then are removed from chromatin after replication ., For example , the DNA replication factor PCNA is found associated with nascent DNA immediately following the EdU pulse but is largely absent following the 90 minute thymidine chase while the levels of histone H3 and H4 remain constant ( Figure 6A , right panel ) ., It is important to note that the levels of PCNA and histones H3 and H4 do not vary between the Hat1+/+ and Hat1−/− MEFs indicating that the rate of EdU incorporation is not altered by the loss of Hat1 ., As previously reported , the levels of histone H4 lysine 5 and lysine 12 acetylation are high on nascent DNA and then decrease over the 90 minute thymidine chase in the Hat1+/+ MEFs 48 ., In the Hat1−/− cells there is a striking decrease in the levels of H4 lysine 5 and 12 acetylation on the nascent DNA ., The low level of acetylation that remains does not decay over time and is consistent with the observation that parental histones can remain acetylated during their reassembly during DNA replication 50 ., These results indicate that Hat1 is likely to be the sole histone acetyltransferase involved in the acetylation of histone H4 lysines 5 and 12 during replication coupled chromatin assembly ., We also examined the dynamics of histone H3 modification during replication-coupled chromatin assembly ., In these experiments cells were given either 10 or 30 minute pulses of EdU ., Following the 30-minute EdU pulse , the cells were given a thymidine chase for 10 , 60 or 120 minutes ., Surprisingly , examination of the input samples indicated that loss of Hat1 impacted the steady state levels of acetylation at specific H3 lysine residues ( Figure 6B , left panel ) ., When normalized to unmodified histone H3 , there was a marked decrease in the overall abundance of acetylated H3 lysine 9 ( >2-fold ) and a moderate decrease in the acetylation of lysines 18 and 27 ( <2-fold ) ., In addition , there was an increase in the level of H3 lysine 14 acetylation ( ∼2-fold ) ., Importantly , Hat1 also had a significant influence on the dynamics of histone H3 acetylation on nascent DNA ., In both Hat1+/+ and Hat1−/− cells , total histone H3 levels increased during the pulse , as more nascent DNA was labeled , and then remained constant throughout the chase ( Figure 6B , right panel ) ., As expected , PCNA levels increased during the EdU pulseand then decreased during the thymidine chase 48 ., Interestingly , distinct patterns of acetylation dynamics were observed for subsets of the lysine residues on the NH2-terminal tail of histone H3 ., The acetylation of H3 lysines 14 and 23 largely mirrored that of bulk H3 where the levels remained relatively stable during the thymidine chase ., This acetylation was not significantly influenced by the loss of Hat1 ., A second pattern , seen for lysines 9 , 18 and 27 , displayed kinetics similar to those seen for H4 lysine 5 and 12 acetylation where acetylation increased during the EdU pulse and then decayed to a basal level during the thymidine chase ., Surprisingly , the acetylation of lysines 9 , 18 and 27 was sensitive to the loss of Hat1 and showed only a basal level of acetylation that did not decay during chromatin maturation ., Therefore , in addition to its expected effects on histone H4 , the presence of Hat1 is also essential for acetylation of histone H3 deposited during replication-coupled chromatin assembly ., In addition to acetylation , newly synthesized histone H3 is also monomethylated on lysine 9 ., In fact , H3 lysine 9 monomethylation appears to be precede any other modifications on H3 and H4 43 , 51 ., H3 lysine 9 monomethylation was apparent on nascent DNA and then increased during the course of the thymidine chase ( Figure 6B ) ., The levels and kinetics of H3 lysine 9 monomethylation were not influenced by the absence of Hat1 ., Hence , the mono-methylation of newly synthesized histone H3 , which is thought to occur prior to its association with the Hat1 complex , is not dependent on Hat1 ., The effect of Hat1 on the acetylation state of histones incorporated during replication-coupled chromatin assembly suggested that Hat1 is modifying newly synthesized molecules ., To test this , Hat1+/+ and Hat1−/− MEFs were briefly pulsed with 3H-lysine to radiolabel newly synthesized proteins ., Histones were then purified from these cells by acid extraction and resolved by acid-urea ( AU ) gel electrophoresis ., AU gels are capable of resolving the acetylated isoforms of histones where the addition of each acetyl groups causes a successive decrease in electrophoretic mobility ., The AU gels were stained with coomassie blue and then processed for fluorography ( Figure 7 ) ., The coomassie blue stained gel shows the mobility and distribution of bulk histones ., The absence of Hat1 had little effect on the bulk histones ., Examining the radiolabeled histones provides specific information on the distribution of acetylated isoforms of the newly synthesized histones ., In Hat1 WT cells , essentially all of the newly synthesized histone H4 migrated at a position consistent with the diacetylated state , in agreement with previous reports ., However , in the absence of Hat1 , it appeared that nearly all of the newly synthesized histone H4 was found to be unacetylated ., This conclusively demonstrates that Hat1 is involved in the acetylation of newly synthesized histone H4 and appears to be the only enzyme responsible for this pattern of acetylation ., Surprisingly , loss of Hat1 also altered the distribution of newly synthesized histone H3 ., While histone H3 is more difficult to resolve in AU gels , newly synthesized histone H3 isolated from Hat1 WT cells showed a distribution of isoforms ., In the absence of Hat1 , there was a decrease in the modified isoforms and a marked increase in unacetylated newly synthesized histone H3 ., This is consistent with the effect of Hat1 on the acetylation state of H3 deposited during replication-coupled chromatin assembly and suggests the possibility that the proper processing of newly synthesized histone H3 is linked to processing and acetylation of newly synthesized histone H4 ., Hat1 function has been studied in a number of model organisms , including S . cerevisiae , S . pombe and chicken DT40 cells ., The absence of Hat1 in these organisms did not have any significant impact on overall cell proliferation 20 , 21 , 34 , 37 , 38 ., Combined with the absence of a significant phenotype arising from mutating histone H4 lysine 5 and 12 in budding yeast had led to the idea that the evolutionarily conserved diacetylation pattern on newly synthesized histone H4 is either not involved in chromatin assembly or plays only an accessory or specialized role in this process 52 , 53 , 54 ., The current analysis of mammalian Hat1 indicates that , while Hat1 is essential for viability in the mouse , it is not essential for cell proliferation ., Indeed , the loss of viability seen in the Hat1−/− neonates appears to be the result of specific developmental defects that result in cellular hyperproliferation ., It should be stressed that the direct cause of the morphological defects observed in the Hat1−/− neonates is not known ., However , potential effects on development due to alterations in chromatin assembly are consistent with the recent report that a mutation in the replication-coupled histone H3 variant , H3 . 1 , or mutations in the histone chaperone CAF-1 cause specific neural development defects in C . elegans 55 ., Hence , while the current study does not address whether the essential function of Hat1 is related to its impact on histone deposition or through an as yet unidentified cellular role , the use of developmentally complex organisms may facilitate our understanding of the in vivo consequences of manipulating chromatin assembly pathways ., A function for Hat1 in DNA damage repair appears to be evolutionarily conserved ., However , Hat1 appears to play a more extensive role in mammalian cells than in the other organisms examined ., For example , deletion of HAT1 in S . cerevisiae ( when combined with specific mutations in histone H3 ) results in sensitivity to MMS and to the exogenous expression of restriction endonucleases 36 ., Likewise , loss of Hat1 in S . pombe and chicken DT40 cells increases sensitivity to MMS ., However , loss of Hat1 in these organisms does not increase sensitivity to UV exposure suggesting that the role of Hat1 is limited to double strand break repair 37 , 38 ., However , the Hat1−/− MEFs display sensitivity to both double strand and single strand damaging agents ., Importantly , mammalian Hat1 mutants also display profound defects in genome instability , which has not been observed elsewhere ., These observations suggest that mammalian DNA repair-linked chromatin assembly pathways may be more dependant on the proper modification state of newly synthesized histones or that Hat1 may play a more direct and integral role in DNA repair mechanisms in mammalian cells ., In addition , the observation of chromosomal fusions in the absence of Hat1 may reflect a disruption of telomere structure , which is another property of Hat1 mutants that is evolutionarily conserved 56 , 57 ., The process of chromatin assembly is both spatially and temporally dynamic , which has complicated efforts to definitively demonstrate that Hat1 is involved in histone deposition ., The alterations in histone acetylation patterns observed on nascent DNA in Hat1−/− MEFs directly links Hat1 to replication-coupled chromatin assembly ., Combined with two other recent reports , the Hat1 enzyme has now been directly linked to the process of chromatin assembly in three distinct contexts ., Consistent with the DNA double strand break sensitivity of hat1Δ budding yeast , the absence of Hat1 resulted in defects in the reassembly of chromatin structure that is linked to the recombinational repair of DNA double strand breaks 39 ., In addition , a genetic screen in yeast identified Hat1 as a factor important for replication-independent chromatin assembly ( or histone exchange ) 40 ., These observations suggest that the acetylation of newly synthesized histones is a ubiquitous feature of all chromatin assembly pathways consistent with the presence of the lysine 5 and 12 diacetylation pattern on both histones H3 . 1 and H3 . 3 43 , 50 , 51 , 58 , 59 , 60 ., While it is clear that Hat1 is involved in chromatin assembly , the precise function of the lysine 5 and 12 diacetylation pattern on newly synthesized histone H4 has not been identified ., Recent reports have suggested that this diacetylation pattern promotes the nuclear import of new H4 perhaps through increasing interactions with Importin 4 42 , 51 ., However , the analysis of histone dynamics on nascent DNA presented here suggest that in the absence of diacetylation on newly synthesized H4 , the level and kinetics of H3 and H4 deposition is similar ., This suggests that any impact on nuclear import is not critical for histone deposition during DNA replication in mammalian cells ., The iPond analysis of the acetylation state of histone H3 during replication-coupled chromatin assembly indicated that histone deposition is accompanied by the transient acetylation of histone H3 lysines 9 , 18 and 27 ., The fact that decay of these acetylations is similar to that of H4 lysine 5 and 12 acetylation and that these acetylations are dependent on Hat1 suggests that this acetylation pattern may be a hallmark of chromatin assembly in murine cells ., However , this pattern of acetylation does not match those observed on soluble histones in other mammalian systems ., For example , newly synthesized histone H3 . 1 in HeLa cells appears to be unacetylated while newly synthesized H3 . 2 and H3 . 3 showed low levels of acetylation on all of the lysine residues in the H3 NH2-terminal tail 50 ., In addition , soluble histones H3 . 1 and H3 . 3 from HeLa cells showed low levels of acetylation on lysine 9 and moderate levels of acetylation on either lysine 14 or 18 ( lysine 27 was not examined ) 58 , 6 | Introduction, Results, Discussion, Materials and Methods | Histone acetyltransferase 1 is an evolutionarily conserved type B histone acetyltransferase that is thought to be responsible for the diacetylation of newly synthesized histone H4 on lysines 5 and 12 during chromatin assembly ., To understand the function of this enzyme in a complex organism , we have constructed a conditional mouse knockout model of Hat1 ., Murine Hat1 is essential for viability , as homozygous deletion of Hat1 results in neonatal lethality ., The lungs of embryos and pups genetically deficient in Hat1 were much less mature upon histological evaluation ., The neonatal lethality is due to severe defects in lung development that result in less aeration and respiratory distress ., Many of the Hat1−/− neonates also display significant craniofacial defects with abnormalities in the bones of the skull and jaw ., Hat1−/− mouse embryonic fibroblasts ( MEFs ) are defective in cell proliferation and are sensitive to DNA damaging agents ., In addition , the Hat1−/− MEFs display a marked increase in genome instability ., Analysis of histone dynamics at sites of replication-coupled chromatin assembly demonstrates that Hat1 is not only responsible for the acetylation of newly synthesized histone H4 but is also required to maintain the acetylation of histone H3 on lysines 9 , 18 , and 27 during replication-coupled chromatin assembly . | The packaging of genomic DNA during replication is a highly orchestrated process ., An important aspect of chromatin assembly is the processing of newly synthesized histones prior to their incorporation into chromatin ., The transient acetylation of histone H3 and H4 NH2-terminal tails is a hallmark of this processing with newly synthesized molecules of histone H4 being predominantly diacetylated ., This diacetylation occurs specifically on lysine residues 5 and 12 and this precise pattern is widely conserved throughout eukaryotic evolution ., The acetylation of newly synthesized histones is catalyzed by type B histone acetyltransferases ., Hat1 is the founding member of this class of enzymes and has been proposed to be responsible for the diacetylation of newly synthesized histone H4 ., Here we describe the development of a mouse knockout model of Hat1 ., The absence of Hat1 results in neonatal lethality due to developmental defects in the lung ., Mouse embryonic fibroblasts derived from Hat1−/− mice are sensitive to DNA damaging agents and display a high level of genome instability ., Biochemical analyses provide definitive evidence that Hat1 is the sole enzyme responsible for the acetylation of newly synthesized histone H4 ., Surprisingly , Hat1 is also necessary for the normal processing of newly synthesized histone H3 . | biology | null |
journal.pcbi.1002432 | 2,012 | Feedforward Inhibition and Synaptic Scaling – Two Sides of the Same Coin? | As part of an ever-changing world , brain activity changes continuously ., The fraction of neurons active in a region at each given moment fluctuates significantly driven by changes in the environment and intrinsic dynamics ., Ideally , regions receiving this activity as input should be able to represent incoming signals reliably across the full possible range of stimulation conditions ., Indeed , this type of regulation seems to be ubiquitous in the cortex ., In the early visual system , contrast gain control begins in the retina 1 and is strengthened at subsequent stages of the visual system , such that the way an image is represented in V1 simple cells is largely contrast invariant 2 , 3 ., Similarly , in the olfactory system , neuronal representations remain sparse and odor-specific over thousand-fold changes in odor concentration 4–6 ., To be able to achieve such invariance , neurons have evolved various mechanisms that adjust neuronal response properties as function of their total input ., One instance of such normalization involves feedforward inhibition , in which afferent inputs induce both excitation and mono-synaptically delayed inhibition onto principal cells 7–12 , shaping the temporal activity pattern of the postsynaptic neurons 8–10 , and sparsifying population activity 5 ., The degree of specificity of this inhibition can vary from stimulus specific to relatively unspecific 7 , 12 ., Here , we focus on fast but unselective feedforward inhibition , which has been reported in a range of circuits including hippocampus and sensory areas 11 , 13–15 ., This mechanism adjusts , virtually instantaneously , the sensitivity of pyramidal cells to the overall strength of the afferent stimulus ., As a result , the influence of an individual afferent on the firing of the postsynaptic neuron is continuously normalized by the total number of active afferents ., Functionally , it has been hypothesized that such input normalization is needed to expand the range of inputs that can be represented in a neuron population 11 , however , its implications for learning in the circuit remain unclear ., Another mechanism with similar effects , but acting on a slower time scale , is synaptic scaling 16–18 ., Specifically , it is believed that neurons detect sustained changes in their firing rates through calcium-dependent sensors and increase or decrease the density of glutamate receptors at synaptic sites to compensate for these changes in drive 19 ., This results in an uniform rescaling of the strength of excitatory synapses as a function of average postsynaptic activity ., Synaptic scaling often takes a multiplicatively form 17 , which has the benefit of preserving the relative contribution of synapses and hence the information stored through Hebbian learning 20 ., This type of weight normalization is believed to address a different kind of stability problem–the fact that synapses are plastic ., As Hebbian learning alone would destabilize neural dynamics , due to a positive feedback loop , additional homeostatic mechanisms such as synaptic scaling are needed to ensure stable circuit function 18–20 ., Fast feedforward inhibition and synaptic scaling have been reported for a range of circuits including hippocampal and neocortical pyramidal neurons 11 , 19 ., Given that both mechanisms effectively regulate the total incoming drive to neurons , it may be somewhat surprising that they co-occur in the same cell types ., This suggests there may be some computational advantage in combining input normalization and synaptic scaling ., However , based on the existing experimental evidence alone , it is unclear what possible benefits this interaction may have ., We show here that the role of input normalization and synaptic scaling goes beyond simply maintaining circuit homeostasis , and that they play important computational roles during synaptic learning ., In the presence of neuronal competition by global lateral inhibition , the two enable efficient unsupervised learning from noisy or ambiguous inputs ., Specifically , we consider an elementary circuit that incorporates synaptic scaling and fast feedforward inhibition ., We analyze the learning dynamics in this circuit and show that , for certain input statistics , standard neural dynamics and Hebbian synaptic plasticity implement approximately optimal learning for this data–an observation that we further confirm in numerical experiments ., The studied circuit learns an efficient representation of its inputs which can be used for further processing by downstream networks ( e . g . , for classification ) ., Importantly , in the absence of feedforward inhibition , learning in the same circuit results in much poorer representations , as the system has a stronger tendency to converge to locally optimal solutions–a problem that neural and non-neural systems for unsupervised learning commonly face ., This provides evidence for synaptic plasticity requiring normalized inputs for efficient learning ., Given that feedforward inhibition and synaptic scaling seem to co-occur in various neural circuits , our results suggest that the interplay between the two mechanisms may generally facilitate learning in the cortex ., As a starting point , consider the elementary neural circuit shown in Fig . 1A ., The network consists of neurons receiving excitatory inputs from input neurons through a set of excitatory weights , ., We denote by the activity of input neuron and by the activity of the downstream processing neuron ., In the general case , the activity of neurons can be defined as a function of the activity of the input layer , , and of the weights : ( 1 ) This transfer function is not necessarily local , as it does not restrict the dependency to the afferent weights of neuron ; it allows us to also describe the interactions between neurons through lateral connections ( marked by dotted lines in Fig . 1A ) ., For the first part of the analysis , we assume the neural dynamics given by ( 1 ) to be arbitrary , though later we consider specific forms for the transfer function ., We model feedforward inhibition by explicitly normalizing the input vector to satisfy the constraint: ( 2 ) Such input normalization can remove undesired patterned variations ( e . g . contrast , see Fig . 1D ) , potentially facilitating learning in the circuit ., If we denote the un-normalized input by , the constraint can , for instance , be fulfilled by a simple division , , though alternative implementations are possible ., This formulation abstracts away the details of the biological implementation , focusing on its functional implications 11 ., Importantly , the simple form allows us to derive theoretical results about the role of this form of feedforward inhibition during learning ., At the level of the neural circuit , however , input normalization relies on the presence of a set of fast spiking interneurons ( in the hippocampus – predominantly basket cells 8 ) innervated by the same afferent inputs , with unspecific projections onto the subsequent layer ., The implications of this neural implementation are considered in more detail in the Discussion ., We model incoming synapses to be plastic and to change by Hebbian learning , with synaptic scaling implemented by an additional weight dependent term 20 , 21: ( 3 ) where is a small positive learning rate ., This synaptic scaling model captures the important biological constraint that weight changes should rely only on information that is local to the synapse ., It differs from global forms that use an explicit weight normalization in that the normalizing constant is not a separate parameter , but rather is implicitly determined by the circuit dynamics ., The circuit model above defines specific learning dynamics for the synaptic weights as function of the their initial values and of the incoming inputs ., To investigate the evolution of the weights analytically , it is informative to first study the time course of the weight sums for an arbitrary neuron ., Using the learning rule ( Eq . 3 ) and the explicit input normalization constraint ( Eq . 2 ) , we obtain: ( 4 ) which shows that is a stationary point for the dynamics of ., Furthermore , since neural activity and the learning rate are both positive , is a stable stationary point , i . e . , increases when smaller than , and decreases when larger , independent of the input statistics ., Consequently , synaptic plasticity automatically adjusts the sum of the incoming weights to each neuron to the total incoming drive ( since ) ., Hence , the synaptic weights of a processing neuron adapt during learning to match the scale of its inputs ., Rather than being a separate parameter , the norm of the weights is inherited from the properties of the input stimuli ., We show below that this match of the normalizing constants for inputs and weights , respectively , is critical for achieving efficient learning in the neural circuit ., In contrast to the mean , which is independent of the inputs provided that the inputs are normalized , the stationary points for individual weights depend on the statistics of the input patterns ., Such a dependency is , of course , needed if the circuit is to memorize properties of the input after learning ., We can derive an analytical solution for learning in this system , something that has often proved difficult for other models ., Specifically , if we consider the input vectors to be drawn independently and identically from a stationary but otherwise unspecified distribution , we can show ( see Methods ) that , at convergence , the weights associated with each neuron are uniquely determined by the statistics of input stimuli and the transfer function : ( 5 ) where the brackets denote the average of the expression under the input distribution ., This approximation is very accurate for small learning rates and large numbers of inputs ., Although Eq ., 5 gives a formal description for the outcome of learning in the neural circuit as a function of the neuron dynamics and the input statistics , it tells us little about the quality of the learning result ., For this , we need to specify the input distribution ., In particular , we use a generative model , which gives not only an explicit model for the input statistics , but also an expression for the theoretically optimal solution for inference and learning on such data , which we can use to evaluate the quality of learning in the neural circuit 22 ., The specific generative model we chose is a mixture model , which is naturally associated with classification tasks 23 ., Intuitively , a mixture model assumes each input stimulus to belong to one out of classes ., Each class is described by a representative input and its typical variations ., Mixture models have been well-investigated theoretically and are used to model a variety of data 23 ., Moreover , although they may seem restrictive , mixtures are well-suited to model multi-modal data distributions also when the assumptions of the model are not satisfied exactly 23 ., In generative model terminology , mixture distributions assume an input to be generated by one of model classes ( see Fig . 1B ) ., Each class is described by a representative pattern , which we will refer to as its generative field ., The mixture distributions define the variations of the patterns within each class , where is the matrix of all generative fields ., The prior probability specifies how many inputs are generated by the different classes ., Here , we assume all classes to be equally likely , and , since inputs represent positive firing rates , we choose the Poisson distribution to model noise: ( 6 ) where is the number of input dimensions ., To capture the effects of feedforward inhibition , we assume the parameters to satisfy the constraint: ( 7 ) with parameter effectively determining the contrast of the inputs , see Fig . 1C ., Note that this model only approximates the effect of feedforward inhibition , since individual stimuli are not normalized ( the constraint in Eq . 2 is only true on average ) ., However , the approximation gets increasingly accurate with increasing size of the stimuli , ., Having a model for the input distribution , we can derive the optimal solution for inference and learning on this data ., In particular , we use the expectation maximization ( EM ) framework 24 , 25 which enables us to learn the maximum likelihood solutions for the parameters from input stimuli ., Intuitively , this optimal learning procedure alternates between what we call the E-step , estimating how likely the data are under the current model , and the M-step , when we change the model parameters ., Iterating E- and M-steps is guaranteed to never decrease the data likelihood and , in practice , it increases the likelihood to ( possibly local ) likelihood maxima ., If a global maximum likelihood solution is found , the parameters represent the best possible learning result ( in the limit of many data points ) ., Similarly , the posterior distribution with optimal represents the best possible inference given any specific input ., For our model , we obtain the following update rules for optimal parameter learning: ( 8 ) ( 9 ) where the posterior probability required for the E-step takes the form of the well-know softmax function 26 with arguments ., With the concrete model of normalized input data , we can now ask how learning in our neural circuit is related to the theoretically optimal solutions for such data ., First , recall that after learning in the neural circuit has converged , the synaptic weights are a solution of Eq ., 5 ., Second , for the probabilistic model the ( possibly local ) optimum is obtained after the EM iterations have converged , which means that satisfies Eq ., 9 with ., Comparing the result of neural learning with the result of EM learning , we note that they have a very similar structure: ( 10 ) Indeed , synaptic weights can be easily mapped into the parameters of the generative model and if we choose the transfer function in the circuit to be equal to the posterior probability , the two expressions are the same ., Hence , if we interpret neural activity as representing posterior probabilities under our model ( compare 27–31 ) , any fixed point of EM optimization becomes an approximate fixed point of neural learning ., The transfer function makes learning in the neural circuit approximately optimal for normalized data , but what does this transfer function mean in neural terms ?, First , the optimal neural dynamics requires a specific form of lateral interactions , implementing the softmax function ( Eq . 8 , left-hand-side ) ., Through these interactions , neurons compete for representing each input stimulus ., Due to its importance for competitive learning , neural circuits giving rise to the softmax have extensively been investigated 26 , 32–34 ., Typically they involve unspecific feedback inhibition which suppresses neurons with weak inputs while those with strong inputs can maintain high activity rates ., Most of the variants of the implementation should work for the purposes of our model ( also compare 35–37 ) ; hence we do not commit to one specific realization of this function ., The arguments of the softmax have a particularly simple form: they represent local summations of input activities weighted by synaptic strengths , ., While the summation of inputs is biologically plausible , scaling by the logarithm of the weights may not be ., It , for instance , implies that the contribution of an input to a neurons activity may be negative or , unrealistically , change sign during learning ., This problem can be addressed , however , while preserving the close correspondence between the circuits fixed points and maximum likelihood solutions ., To achieve this , we note that the only requirement for the input data is that the total input is preserved , ., We therefore have some freedom when modeling how feedforward inhibition enforces this constraint ., In particular , if the un-normalized input is , then feedforward inhibition could constrain the total inputs by: ( 11 ) which represents a slight alteration to the common choice ., Practically , this form of normalization continues to scale the activity of an un-normalized input unit by the total activity , but it introduces an offset corresponding to having some spontaneous background activity in the input layer ( which leads to a normalization constant ) ., This model of feedforward inhibition guarantees that the weights will eventually converge to values larger or approximately equal to one ., As a consequence , negative weight factors can be removed completely by linearizing the logarithm around one ., We consider two forms of such a linearization: in the first , we use the linearization only for values of , in the second , we completely replace the logarithm by the linearized form ( see inset of Fig . 1A ) : ( 12 ) where ., For the linearization we exploited that for normalized inputs the softmax becomes invariant with respect to weight offsets ( see Methods ) ., The linear case recovers the conventional linear summation of synaptic inputs , while the logarithmic case is a closer approximation of the optimal dynamics ( see Discussion ) ., The complete description of the final neural circuit is summarized in Table 1 ., It consists of essentially three elements: input normalization , Hebbian plasticity with synaptic scaling , and softmax competition ( see also Fig . 1 ) ., Our analysis shows that these elementary models of neural interactions can be approximately optimal for learning on normalized inputs from mixture distributions ., Notably , the neural circuit can process any type of un-normalized data as feedforward inhibition projects any stimulus to a subspace on which learning is optimal ., It is important to remark that no explicit knowledge about is required at the level of processing neurons , which would be difficult to justify neurally ., Instead , synaptic scaling automatically adjusts the weights such that the constraint in Eq ., 2 is satisfied ., This , furthermore , means that synaptic plasticity can follow slow changes of the normalization constant , which could be used to further facilitate learning ., Formally , manipulating during learning provides a simple implementation for simulated annealing , which is often used to prevent optimization from converging to locally optimal solutions 38 , 39 ., Alternatively , annealing can be achieved by changing the amount of spontaneous activity in the input layer ( see Discussion for neural mechanisms implementing such changes ) ., Considering the details of the neural circuit and the generative model used here , some aspects of the analytical results presented may not seem very surprising ., The similarity between the fixed points for the synaptic weights and the maximum likelihood solution is partly due to the fact that both models fulfill the same constraint , , at least approximately ., However , this constraint has different origins in the two models: in the neural circuit it is a reflection of synaptic scaling , whereas in the generative model it appears due to the fact that the modeled data is normalized ., Along the same lines , the fact that the softmax function emerges as the optimal transfer function for the circuit is somewhat expected , given that the softmax is closely associated with mixture models ., However , the arguments of the softmax , , have a particularly compact form in our case , and they can be easily approximated through the integration of afferent inputs to the processing neurons ., The compactness of the neural interactions is a direct consequence of the combination of Poisson mixture distributions , normalized inputs and synaptic scaling ., Without any of these components , the interactions would be more complicated , or not optimal ., Although we have shown that learning in the neural circuit approximates optimal learning for our data model , several details remain to be investigated ., First , it is unclear how close is learning in the neural circuit to the optimum in practice ., Second , since real data rarely follows the assumptions of the model exactly , we would like to know how robust learning is in such cases ., These questions can only be answered through numerical simulations using either simple artificial data for which the optimal solutions are known , or realistic inputs from a standard database ., Until now , we have evaluated the effectiveness of learning by measuring how well the final weights can describe the data ( formally , the data likelihood under the generative model ) ., Alternatively , we could ask how useful the emerging input representation is for performing higher level tasks in downstream circuits ., The performance for such tasks can give a measure of learning quality that is more independent of specific assumptions about the input statistics ., Moreover , such alternative performance measures become a necessity when comparing learning on normalized versus un-normalized data , as done in the following section ., Since likelihoods are well-suited measures of learning performance only when computed using the same data , no such comparison is possible when trying to asses the benefits of normalization ., For the MNIST dataset , a natural task is classification , which has been extensively investigated in the literature , both in neural models and using purely functional approaches ( e . g . , 42–44 ) ., Note , however , that the type of classification relevant for biological systems differs from the generic classification in several aspects ., Perhaps most importantly , stimuli processed by neural circuits usually come without explicit labels ., For instance , most visual stimuli we process are not accompanied by labels of the visual objects that caused them ., However , during development we are provided ( directly or indirectly ) with the meaning of objects for some stimuli ., In order to classify inputs accordingly , the model needs to have access to at least some stimuli for which the class membership ( label ) is known ., These labels can then be used to associate the representations in the lower processing layer ( obtained by unsupervised learning ) with the corresponding class; for instance , all writing styles of a hand-written ‘2’ with digit class ‘2’ ., Having an overcomplete representation of the data becomes critical for the system to work in this setup ., As we have seen in previous numerical experiments , learning with MNIST data yields representations of different classes of hand-written digits ., Because of different writing styles , the variations of all patterns showing the same digit are too strong to allow for a representation of all digits with one class per digit ., However , as already shown in Fig . 3D , with more neurons than classes , the emergent representation successfully captures all digit classes , with different neurons representing different writing styles ( the more units , the more detailed the representation of different writing styles ) ., For classification , we extend the neural circuit to include an additional processing stage that makes use of the previously learned representation for assigning class labels ., As done for the first processing layer , we formulate the classification process probabilistically , using a generative model assuming that a digit type generates different writing styles ( Fig . 4A ) ., This allows us to derive a probabilistic procedure for classifying a given input stimulus ( see Methods ) ., The focus here is assessing the utility of the first layer representation for higher level computations rather than the neural implementation of this later processing stage ., Still , we can note that the dynamics of the second layer shares several features with the first layer model: the neural dynamics have a simple dependency on a weighted sum of incoming inputs ( see Methods ) , and the inputs themselves are normalized ( because of the softmax ) , suggesting this type of computation could be implemented in a neurally plausible circuit ., To illustrate classification based on the representations learned unsupervised , we first consider stimuli representing digits of types ‘0’ to ‘3’ ., For this data , the representations learned by unsupervised learning in the first processing layer ( with units ) is shown in Fig . 4B ( bottom row ) ., We label these representations using of the data used for training ( i . e . , we use the labels of of the training data ) ., The probability distribution for the map between first layer representations and class labels is shown in Fig . 4B ( computed using Eq . 34 , see Methods ) , demonstrating a close to perfect assignment of representations to digit classes ., For a quantitative analysis of this match , we can measure the classification performance of the system for a test dataset ( i . e . , for data not used for training; see Methods for details ) ., For the four digit dataset , the classification performance as function of the number of neurons in the first processing layer is shown in Fig . 4C ., For both the neural circuit and EM optimization classification performance increases with the number of units ., As can be observed , the neural circuit with log-saturating synaptic efficacies shows virtually identical classification rates to EM learning ., Likewise , the neural circuit with standard linear input summation shows a good classification performance , even slightly better for the complete case ( four digit classes and four processing neurons ) ., In an overcomplete setup , the rate of successful classifications is still high ( e . g . , around for the five times overcomplete setup ) , though a bit lower than for the log case and EM ., So far , we have used classification performance as an additional measure for the quality of learning in the circuit ., However , the setup is interesting from a functional perspective as well , since it allows for relatively high rates of correct classification using a very limited amount of labeled data ., Fig . 4E shows classification performance for different degrees of overcompleteness in the processing layer if normalized EM is applied to the full MNIST data ( we use EM here as it can be efficiently scaled-up to the size of the full MNIST dataset; see Methods ) ., As before , classification performance increases with an increasing number of units and with the number of labels used for classification ( see Fig . 4E and Fig . 4F , respectively ) ., Importantly , a small percentage of labels is already sufficient to obtain almost the same classification performance as when using all labels ., For instance for processing units we obtained a performance of correctly classified stimuli using just of the MNIST labels ., For rates above less than of labels were sufficient ., Moreover , performance in our model is comparable to that of state-of-the-art methods , such as deep belief networks ( DBN; 42 ) ., Using all the labels , the performance of DBN reaches 42 , but with a much more complex circuit ( two processing layers and an associative memory ) , several learning mechanisms , and after the tuning of many free parameters ., In contrast , learning in our model is very straightforward , with very few free parameters ( ) , and requires just few labeled inputs ., These properties seem particularly desirable in the biologically relevant setting ., Even if we assume that synaptic scaling is unavoidable to guarantee stability during Hebbian learning , it is still unclear why the system would need feedforward inhibition , or , more in formal terms , what are the benefits of learning using normalized data ., This question can be addressed at two levels ., First , at an abstract level , we can ask how different are the outcomes of optimal probabilistic learning when using unconstrained versus normalized data ., Second , in neural terms , we can ask how learning changes when blocking feedforward inhibition in the neural circuit ., To answer the first question , we use our generative model approach to compare the optimal learning dynamics for data that is , or not , normalized ( this difference will depend on the relative size of different stimuli; compare Fig . 5A and B ) ., Formally , we construct an analog mixture model for un-normalized data , and derive optimal learning for this model ., The analysis yields a similar set of update rules ( see Methods , Eqs . 26 and 27 ) , which we can use for unsupervised learning with similar ( but un-normalized ) data ., Because the two learning procedures use different data , comparing them is nontrivial ., While for data generated according to the assumed probabilistic model we can still use the percentage of trials converging to the optimum as a performance measure , comparison becomes very difficult for the digits data ., Since the likelihoods are no longer comparable ( because they are estimated from different data ) , we can only rely on the classification rates for estimating the quality of the learned representations in this case ., We compare the performance of the two learning procedures for the same two datasets described above ., For the blocks dataset , learning performance is not significantly different in the two cases ( not shown ) , probably because the task is too easy to be able to differentiate between the two learning procedures ., The results for the digits are shown in Fig . 5C ., The unconstrained learning procedure yields worse performance than the constrained case; the difference may seem small in absolute terms , but the classification rate for the unconstrained case is worse than the outcome of k-nearest-neighbour ( k-NN ) classification , which we may view as a lower bound for task difficulty ., In itself , this result is not sufficient to prove that learning from normalized data is generally useful for unsupervised learning ., Since we can only estimate learning performance indirectly , through the classification rates , it may be that data normalization improves classification in general , by removing task irrelevant variability , without having any specific benefit for learning per se ., If this were the case , then we should observe a similar performance improvement for the normalized relative to the unnormalized data when using a standard classifier , such as k-NN ., This is however not the case; on the contrary , for k-NN performance decreases to ( from ) after data normalization , suggesting that the benefits of normalization are restricted to learning procedures that explicitly exploit this property , as does learning in our model ., For the neural circuit , the utility of the interaction between feedforward inhibition and synaptic scaling is further emphasized ., When blocking feedforward inhibition ( practically , this means using unnormalized stimuli as inputs to the circuit ) the linear circuit converges to represent all classes very rarely , much less often than when feedforward inhibition is active in the circuit ( Fig . 5D , compare grey and red bars ) ., In principle , since the neural circuit approximatively implements optimal learning for normalized data , one could expect that performance should be similar to that obtained by constrained EM with un-normalized data , which is indistinguishable from that obtained when learning from normalized data ., So why is there a the big difference in performance in the case of the neural circuit ?, The critical difference between EM and the network is that synaptic scaling only enforces the constraint of the weights through its ( normalized ) inputs ., If the incoming stimuli are not normalized , the sum of the weights is not guaranteed to converge at all ( Eq . 4 does not apply ) ., This intuition is confirmed by the fact that when replacing synaptic scaling by an explicit weights normalization ( see Methods ) learning evolves similarly to the case when feedforward inhibition is active ., These results suggest that feedforward inhibition is critical for correctly learning the structure of the data when the weig | Introduction, Results, Discussion, Methods | Feedforward inhibition and synaptic scaling are important adaptive processes that control the total input a neuron can receive from its afferents ., While often studied in isolation , the two have been reported to co-occur in various brain regions ., The functional implications of their interactions remain unclear , however ., Based on a probabilistic modeling approach , we show here that fast feedforward inhibition and synaptic scaling interact synergistically during unsupervised learning ., In technical terms , we model the input to a neural circuit using a normalized mixture model with Poisson noise ., We demonstrate analytically and numerically that , in the presence of lateral inhibition introducing competition between different neurons , Hebbian plasticity and synaptic scaling approximate the optimal maximum likelihood solutions for this model ., Our results suggest that , beyond its conventional use as a mechanism to remove undesired pattern variations , input normalization can make typical neural interaction and learning rules optimal on the stimulus subspace defined through feedforward inhibition ., Furthermore , learning within this subspace is more efficient in practice , as it helps avoid locally optimal solutions ., Our results suggest a close connection between feedforward inhibition and synaptic scaling which may have important functional implications for general cortical processing . | The inputs a neuron receives from its presynaptic partners strongly fluctuate as a result of either varying sensory information or ongoing intrinsic activity ., To represent this wide range of signals effectively , neurons use various mechanisms that regulate the total input they receive ., On the one hand , feedforward inhibition adjusts the relative contribution of individual inputs inversely proportional to the total number of active afferents , implementing a form of input normalization ., On the other hand , synaptic scaling uniformly rescales the efficacy of incoming synapses to stabilize the neurons firing rate after learning-induced changes in drive ., Given that these mechanisms often act on the same neurons , we ask here if there are any benefits in combining the two ., We show that the interaction between the two has important computational consequences , beyond their traditional role in maintaining network homeostasis ., When combined with lateral inhibition , synaptic scaling and fast feedforward inhibition allow the circuit to learn efficiently from noisy , ambiguous inputs ., For inputs not normalized by feed-forward inhibition , learning is less efficient ., Given that feed-forward inhibition and synaptic scaling have been reported in various systems , our results suggest that they could generally facilitate learning in neural circuits ., More broadly , our work emphasizes the importance of studying the interaction between different plasticity mechanisms for understanding circuit function . | biology, neuroscience | null |
journal.pgen.1006288 | 2,016 | Using Genetic Distance to Infer the Accuracy of Genomic Prediction | Predicting unobserved phenotypes using high-density SNP or sequence data is the foundation of many applications in medical diagnostics 1–3 , plant 4 , 5 and animal 6 breeding ., The accuracy of genomic predictions will depend on a number of factors: relatedness among genotyped individuals 7 , 8; the density of the markers 7 , 9 , 10; and the genetic architecture of the trait , in particular the allele frequencies of causal variants 11 , 12 and the distribution of their effect sizes 7 ., Most of these issues have been explored in the literature , and have been tackled in various ways either from a methodological perspective or by producing larger data sets and more accurate phenotyping ., However , the extent to which predictive models generalise from the populations used to train them to distantly related target populations appears not to have been widely investigated ( two exceptions are 7 , 13 ) ., The accuracy of prediction models is often evaluated in a general setting using cross-validation with random splits , which implicitly assumes that test individuals are drawn from the same population as the training sample; in that case accuracy to predict phenotypes is only bounded by heritability , although unaccounted “missing heritability” is common 14 , 15 ., However , this assumption is violated in many practical applications , such as genomic selection , that require predictions of individuals that are genetically distinct from the training sample: for instance , causal variants may differ in both frequency and effect size between different ancestry groups ( in humans , e . g . 16 for lactose persistence ) , subspecies ( in plants and animals , e . g . 17 for rice ) or even families 18 ., In such cases cross-validation with random splits may overestimate predictive accuracy due to the mismatch between model validation and the prediction problem of interest 19 , 20 even when population structure is taken into account 21 ., The more distantly the target population is related to the training population , the lower the average predictive accuracy of a genomic model; this has been demonstrated on both simulated and real dairy cattle data 20 , 22 , 23 ., In this paper we will investigate the relationship between genetic distance and predictive accuracy in the prediction of quantitative traits ., We will simulate training and target samples with varying genetic distances by splitting the training population into a sequence of pairs of subsets with increasing genetic differentiation ., We will measure predictive accuracy with Pearson’s correlation , which we will estimate by performing genomic prediction from one subset to the other in each pair ., Among various measures of relatedness available in the literature , we will consider mean kinship and FST , although we will only focus on the latter ., We will then study the mean Pearson’s correlation as a function of genetic distance , which we will refer to as the “decay curve” of the former over the latter ., This approach is valuable in addressing several key questions in the implementation of genomic selection programs , such as: How often ( e . g . , in terms of future generations ) will the genomic prediction model have to be re-estimated to maintain a minimum required accuracy in the predictions of the phenotypes ?, How should we structure our training population to maximise that accuracy ?, Which new , distantly related individuals would be beneficial to introduce in a selection program for the purpose of maintaining a sufficient level of genetic variability ?, A baseline model for genomic prediction of quantitative traits is the genomic BLUP ( GBLUP; 24 , 25 ) , which is usually written as, y = μ + Zg + ε with g ∼ N ( 0 , K σ g 2 ) and ε ∼ N ( 0 , σ ε 2 ) , ( 1 ), where g is a vector of genetic random effects , Z is a design matrix that can be used to indicate the same genotype exposed to different environments , K is a kinship matrix and ε is the error term ., Many of its properties are available in closed form thanks to its simple definition and normality assumptions , including closed form expressions of and upper bounds on predictive accuracy that take into account possible model misspecification 15 ., Other common choices are additive linear regression models of the form, y = μ + X β + ε ( 2 ), where y is the trait of interest; X are the markers ( such as SNP allele counts coded as 0 , 1 and 2 with 1 the heterozygote ) ; β are the marker effects; and ε are independent , normally-distributed errors with variance σ ε 2 ., Depending on the choice of the prior distribution for β , we can obtain different models from the literature such as BayesA and BayesB 25 , ridge regression 26 , the LASSO 27 or the elastic net 28 ., The model in Eq ( 1 ) is equivalent to that in Eq ( 2 ) if the kinship matrix K is computed from the markers X and has the form X XT and β ∼ N ( 0 , VAR ( β ) ) 29 , 30 ., In the remainder of the paper we will focus on the elastic net , which we have found to outperform other predictive models on real-world data 31 ., This has been recently confirmed in 32 ., Predictive accuracy is often measured by the Pearson correlation ( ρ ^ ) between the predicted and observed phenotypes ., When we use the fitted values from the training population as the predicted phenotypes , and assuming that the model is correctly specified , ρ ^ 2 coincides with the proportion of genetic variance of the trait explained by the model and therefore ρ ^ 2 ⩽ h 2 , the heritability of the trait ., ( An incorrect model may lead to overfitting , and in that case ρ ^ 2 ⩾ h 2 . ), When using cross-validation with random splits , ρ ^ CV ⩽ ρ ^ and typically the difference will be noticeable ( ρ ^ CV ≪ ρ ^ ) ., However , ρ ^ C V may still overestimate the actual predictive accuracy ρ ^ D in practical applications where target individuals for prediction are more different from the training population than the test samples generated using cross-validation 14 ., This problem may be addressed by the use of alternative model validation schemes that mirror more closely the prediction task of interest; for instance , by simulating progeny of the training population to assess predictive accuracy for a genomic selection program ., This approach is known as forward prediction and is common in animal breeding 19 , 33 ., Another possible choice is the prediction error variance ( PEV ) ., It is commonly used in conjunction with GBLUP because , for that model , it can be estimated ( for small samples ) or approximated ( for large samples ) in closed form from Henderson’s mixed model equations 34 ., In the general case no closed form estimate is available , but PEV can still be derived from Pearson’s correlation 35 for any kind of model as both carry the same information:, PEV = ( 1 - ρ ^ 2 ) * VAR ( y ) ., ( 3 ), For consistency with our previous work 31 and with 4 , whose results we partially replicate below , we will only consider predictive correlation in the following ., A common measure of kinship from marker data is average allelic correlation 24 , 36 , which is defined as K = kij with, k i j = 1 m ∑ k = 1 m X ˜ i k X ˜ j k ( 4 ), where X ˜ i k and X ˜ j k are the standardised allele counts for the ith and jth individuals and the kth marker ., An important property of allelic correlation is that it is inversely proportional to the Euclidean distance between the marker profiles Xi , Xj of the corresponding individuals: if the markers are standardised, 2 n - 2 k i j = 2 n - 2 ∑ k = 1 m X ˜ i k X ˜ j k = ∑ k = 1 m X ˜ i k 2 + X ˜ j k 2 - 2 X ˜ i k X ˜ j k = ∑ k = 1 m ( X ˜ i k - X ˜ j k ) 2 ., ( 5 ) This result has been used in conjunction with clustering methods such as k-means or partitioning around medoids ( PAM; 37 ) to produce subsets of minimally related individuals from a given sample by maximising the Euclidean distance 14 , 19 , 38 ., At the population level , the divergence between two populations due to drift , environmental adaptation , or artificial selection is commonly measured with FST ., Several estimators are available in the literature , and reviewed in 39 ., In this paper we will adopt the estimator from 40 , which is obtained by maximising the Beta-Binomial likelihood of the allele frequencies as a function of FST ., F ^ ST then describes how far the target population has diverged from the training population , which translates to “how far” a genomic prediction model will be required to predict ., In terms of kinship , we know from the literature that the mean kinship coefficient k ¯ between two individuals in different populations is inversely related to F ^ ST 41: kinship can be interpreted as the probability that two alleles are identical by descent , which is inversely related to FST which is a mean inbreeding coefficient ., Intuitively , the fact that individuals in the two populations are closely related implies that the latter have not diverged much from the former: if k ¯ is large , the marker profiles ( and therefore the corresponding allele frequencies ) will on average be similar ., As a result , any clustering method that uses the Euclidean distance to partition a population into subsets will maximise their FST by minimising k ¯ ., The simulations and data analyses below confirm experimentally that k ¯ and F ^ ST are highly correlated , which makes them equivalent in building the decay curves; thus we will report results only for F ^ ST ( see Section C in S1 Text ) ., We evaluate our approach to construct decay curves for predictive accuracy using two publicly-available real-world data sets with continuous phenotypic traits , and a third , human , genotype data set ., We estimate a decay curve of ρ ^ D as a function of FST as follows: The pair of subsets produced by k-means corresponds to m = 0 , hence the notation ( F ^ ST ( 0 ) , ρ ^ D ( 0 ) ) , and we increase m by steps of 2 to 20 until the F ^ ST between the subsamples is at most 0 . 005 ., We choose the stepping for each data set to be sufficiently small to cover the interval 0 , F ^ ST ( 0 ) as uniformly as possible ., The larger m is , the smaller we can expect F ^ ST ( m ) to be ., We repeat step 3, ( a ) and 3, ( b ) 40 times for each m to achieve the precision needed for an acceptably smooth curve ., As an alternative approach , we also consider estimating the decay rate of ρ ^ D by linear regression of the ρ ^ D ( m ) against the F ^ ST ( m ) ; we will denote the resulting predictive accuracy estimates with ρ ^ L . For any set value of F ^ ST , we compare the ρ ^ L at that F ^ ST with the corresponding value ρ ^ D from the decay curve estimated by averaging all the ρ ^ D ( m ) for which | F ^ ST ( m ) - F ^ ST | ⩽ 0 ., 01 ., Assuming that the decay curve is in fact a straight line reduces the number of subsamples that we need to generate , enforces smoothness and makes it possible to compute ρ ^ L for values of FST larger than F ^ ST ( 0 ) ., On the other hand , the estimated ρ ^ L will be increasingly unreliable as ρ ^ L → 0 , because the regression line will provide negative ρ ^ L instead of converging asymptotically to zero ., We also regress the ( ρ ^ D ( m ) ) 2 against the ( F ^ ST ( m ) ) 2 to investigate whether they have a stronger linear relationship than the ρ ^ D ( m ) with the F ^ ST ( m ) , as suggested in 22 using simulated genotypes and phenotypes mimicking a dairy cattle population ., The size of the training ( nTR ) and target ( nTA ) subsamples is determined by k-means ., For the data used in this paper , k-means splits the training populations in two subsamples of comparable size; but we may require a smaller nTA ≪ nTR to estimate ρ ^ D ( 0 ) and the ρ ^ D ( m ) while at the same time a larger nTR is needed to fit the genomic prediction model ., In that case , we increase nTR by moving individuals from the target subsample while keeping the F ^ ST ( 0 ) between the two as large as possible ., The impact on the estimated F ^ ST is likely to be small , because its precision depends more on the number of markers than on nTR and nTA 40 ., The estimated ρ ^ D 0 and ρ ^ D ( m ) might be inflated because we are altering the subsets , even when F ^ ST does not change appreciably ., Its variance , which can be approximated as in 49 , decreases linearly in nTA except that can be compensated by generating more pairs of subsamples for each value of m ., We study the behaviour of the decay curves via two simulation studies ., Finally , we estimate the decay curves for some of the phenotypes available in the WHEAT and MICE data ., For both data sets we also produce and average 40 values of ρ ^ CV using hold-out cross-validation ., In hold-out cross-validation we repeatedly split the data at random into training and target subsamples whose sizes are fixed to be the same as those arising from clustering in step 1 of the decay curve estimation ., Then we fit an elastic net model on the training subsamples and predict the phenotypes in the target subsamples to estimates ρ ^ CV ., Ideally , the decay curve should cross the area in which the ( F ^ ST , ρ ^ CV ) points cluster ., The decay curves from the simulations are shown in Figs 1 , 2 and 3 , and the corresponding predictive correlations are reported in Tables 1 and 2 and S1 Text ., The predictive correlations for the WHEAT and MICE data sets are reported in Table 2 , and the decay curves are shown in Figs 1 , 2 and 3 and S1 Text ., A summary of the different predictive correlations defined in the Methods and discussed here is provided in Table 1 ., In all the simulations and the real-world data analyses the ρ ^ D from the decay curve is close to the linear interpolation ρ ^ L; considering all the reference populations in Table 2 and the generation means in Tables A . 1 and A . 2 in S1 Text , | ρ ^ D - ρ ^ L | ≪ 0 ., 02 41 times out of 47 ( 87% ) ., Both estimates of predictive correlation are close to the respective reference values ρ ¯ and ρ ^ P; the difference ( in absolute value ) is ≪ 0 . 05 39 times ( 41% ) and ≪ 0 . 10 69 times ( 73% ) out of 94 ., The proportion of small differences increases when considering only target populations that fall within the span of the decay curve: 23 out of 44 ( 52% ) are ≪ 0 . 05 and 38 are ≪ 0 . 10 ( 84% ) ., This is expected because the decay curve is already an extrapolation from the training population , so extending it further with the linear interpolation ρ ^ L reduces its precision ., Regressing ( ρ ^ D ( m ) ) 2 against the ( F ^ ST ( m ) ) 2 does not produce a stronger linear relationship than that represented by ρ ^ L ( p = 0 . 784 , see Section D in S1 Text ) ., The range of the predictive correlations ρ ^ D ( m ) around the decay curves varies between 0 . 05 and 0 . 10 , and it is constant over the range of observed F ^ ST for each curve ., It does not appear to be related to either the size of the training subsample or the number of causal variants ., This is apparent in particular from the genomic selection simulation , in which both are jointly set to different combinations of values ., Similarly , there seems to be no relationship between the spread and the magnitude of the predictive correlations ( ρ ^ D ( m ) ∈ 0 , 0 . 75 ) ., This amount of variability is comparable to that of other studies ( e . g . , the range of the ρ ^ D ( m ) is smaller than that in the cross-validated correlations in 32 ) once we take into account that the ( F ^ ST ( m ) , ρ ^ D ( m ) ) are individual predictions and are not averaged over multiple repetitions ., Furthermore , subsampling further reduces the size of the training subpopulations; and fitting the elastic net requires a search over a grid of values for its two tuning parameters , which may get stuck in local optima ., Several interesting points arise from the analysis of the real phenotypes in the WHEAT and MICE data , shown in Table 2 and in Figs B . 1 , B . 2 and B . 3 in S1 Text ., Firstly , cross-validation always produces pairs of subsamples with F ^ ST ⩽ 0 ., 01 and high ρ ^ CV that are located at the left end of the decay curve ., The average F ^ ST is 0 . 006 for the WHEAT data and 0 . 001 for the MICE data , and the difference between the average ρ ^ CV and the corresponding ρ ^ D is ≪ 0 . 02 10 times out of 12 ( 83% , see Table B . 4 in S1 Text ) ., The spread of the ρ ^ CV is also similar to that of the ρ ^ D ( m ) ., Secondly , we note that in the WHEAT data all decay curves but that for flowering time cross the 95% confidence intervals for the cross-country predictive correlations ρ ^ P for Germany and UK reported in 4 ., Even in the MICE data , in which all families are near the end or beyond the reach of the decay curves , the latter ( or their linear approximations ) cross the 95% confidence intervals for the ρ ^ P 18 times out of 24 ( 75% ) ., However , we also note that those intervals are wide due to the limited sizes of those populations ., Furthermore , the decay curves for the phenotypes in the WHEAT data confirm two additional considerations originally made in 4 ., Firstly , 4 noted that the distribution of the Ppd-D1a gene , which is a major driver of this flowering time , varies substantially with the country of registration and thus cross-country predictions are not reliable ., Fig B . 1 in S1 Text shows that the decay curve vastly overestimates the predictive correlation for both Germany and the UK ., Splitting the WHEAT data in two halves that contain equal proportions of both alleles of Ppd-D1a and that are genetically closer overall ( F ^ ST = 0 . 04 ) , we obtain a decay curve that fits the predictive correlations reported in the original paper ( ρ ^ D = 0 . 77 , ρ ^ P = 0 . 79 ) ., Secondly , we also split the data according to their year of registration and use the oldest varieties ( pre-1990 ) as a training sample for predicting yield ., Again the decay curve crosses the 95% confidence intervals for the predictive correlations reported in 4 and the correlations themselves are within 0 . 05 of the average ρ ^ D from the decay curve both for 1990-1999 ( F ^ ST = 0 . 028 , ρ ^ D = 0 . 44 , ρ ^ P = 0 . 40 ) and post-2000 ( F ^ ST = 0 . 033 , ρ ^ D = 0 . 44 , ρ ^ P = 0 . 42 ) varieties ., The decay curves from the genomic selection simulation on the original training population ( 200 varieties ) , shown in blue in Fig 1 , span two rounds of selection and three generations ., When considering 200 or 1000 causal variants , the curve overlaps the mean behaviour of the simulated data points ( shown in green ) almost perfectly: the difference between the generation means ρ ¯ and the decay curve is ⩽ 0 . 06 for the first three generations , with the exception of the first generation in the simulation with 1000 variants ( | ρ ¯ - ρ ^ D | = 0 . 09 ) ., As the number of causal variants decreases ( 50 , 10 ) , the decay curve increasingly overestimates ρ ¯ , although the difference remains ⩽0 . 10 for the first two generations; and both show a slower decay than the ρ ¯ ., This appears to be due to a few alleles of large effect becoming fixed by the selection , leading to a rapid decrease of ρ ¯ without a corresponding rapid increase in F ^ ST . The decay curves fitted on the augmented training populations ( 800 varieties , now including those available at the end of the second round of selection , Fig 2 ) fit the first four generations well ( | ρ ¯ - ρ ^ D | ⩽ 0 . 04 for the first two , | ρ ¯ - ρ ^ D | ⩽ 0 . 06 for the third and the fourth ) ., As before , the only exception is the first generation in the simulation with 1000 variants , with an absolute difference of 0 . 09 ., However , the decay curves are also able to capture the long-range decay rates through their linear approximations ., When considering 200 causal variants , | ρ ¯ - ρ ^ L | ≈ 0 ., 08 for generations 5 to 7 and ≈0 . 10 for generations 8 and 9; and | ρ ¯ - ρ ^ L | ≪ 0 ., 05 for generations 4 to 9 when considering 1000 causal variants ., This can be attributed to the increased sample size of the training, population , which both improves the goodness of fit of the estimated decay curve; and makes the decay rate of the ρ ¯ closer to linear , thus making it possible for the ρ ^ L to approximate it well over a large range of FST values ., To investigate this phenomenon , we gradually increased the initial training population to 4000 varieties through random mating and we observed that for such a large sample size ρ ¯ indeed decreases linearly as a function of FST ., We conjecture that this is due to a combination of the higher values observed for ρ ¯ and their slower rate of decay , which prevents the latter from gradually decreasing as ρ ¯ is still far from zero after 10 generations ., In addition , we note that increasing the number of causal variants has a similar effect; with 200 and 1000 causal variants ρ ¯ indeed decreases with an approximately linear trend , which is not the case with 10 and 50 causal variants ., The cross-population prediction simulation based on the HUMAN data ( Fig 3 ) generated results consistent with those above ., As before , the number of causal variants appears to influence the behaviour of the decay curve: while the ρ ^ D ( m ) decrease linearly for 20 , 100 and 2000 casual variants , they converge to 0 . 65 for 5 causal variants ., However , unlike in the genomic selection simulation , the quality of the estimated decay curve does not appear to degrade as the number of causal variants decreases ., This difference may depend on the lack of a systematic selection pressure in the current simulation , which made the decay curve overestimate predictive correlation when considering 10 variants in the previous simulation ., Finally , as in the analysis of the MICE data , the linear approximation ρ ^ L to the decay curve provides a way to extend the reach of the decay curve to estimate predictive correlations ρ ^ P for distantly related populations ( AMERICA , AFRICA , OCEANIA ) ., Again we observe some loss in precision ( see Table 2 ) , but the extension still crosses the 95% confidence intervals of those ρ ^ P 14 times out of 18 ( 78% ) ., Being able to assess the predictive accuracy is important in many applications , and will assist in the development of new models and in the choice of training populations ., A number of papers have discussed various aspects of the relationship between training and target populations in genomic prediction , and of characterising predictive accuracy given some combination of genotypes and pedigree information ., For instance , 51 discusses how to choose which individuals to include in the training population to maximise prediction accuracy for a given target population using the coefficient of determination ., 52 separates the contributions of linkage disequilibrium , co-segregation and additive genetic relationships to predictive accuracy , which can help in setting expectations about the possible performance of prediction ., 53 and 22 link predictive accuracy to kinship in a simulation study of dairy cattle breeding; and 54 investigates the impact of population size , population structure and replication in a simulated biparental maize populations ., The approach we take in this paper is different in a few , important ways ., Firstly , we choose to avoid the parametric assumptions underlying GBLUP and the corresponding approximations based on Henderson’s equations that provide closed-form results on predictive accuracy in the literature ., It has been noted in our previous work 31 and in the literature ( e . g . 32 ) that in some settings GBLUP may not be competitive for genomic prediction; hence we prefer to use models with better predictive accuracy such as the elastic net for which the parametric assumptions do not hold ., Our model-agnostic approach is beneficial also because decay curves can then be constructed for current and future competitive models , since the only requirement of our approach is that they must be able to produce an estimate of predictive correlation ., Secondly , we demonstrate that the decay curves estimated with the proposed approach are accurate in different settings and on human , plant and animal real-world data sets ., This complements previous work that often used synthetic genotypes and analysed predictive accuracy in a single domain , such as forward simulation studies on dairy cattle data ., Finally , we recognise that the target population whose phenotypes we would like to predict may not be available or even known when training the model ., In plant and animal selection programs , one or more future rounds of crossings may not yet have been performed; in human genetics , prediction may be required into different demographic groups for which no training data are available ., Therefore , we are often limited to extrapolating a ρ ^ D to estimate the ρ ^ P we would observe if the target population were available ., Prior information on F ^ ST values is available for many species such as humans 39 , 43; and can be used to extract the corresponding ρ ^ D from a decay curve ., We observe that the decay rate of ρ ^ D is approximately linear in F ^ ST for most of the curves , suggesting that regressing the ρ ^ D ( m ) against the F ^ ST ( m ) is a viable estimation approach ., This has the advantage of being computationally cheaper than producing a smooth curve with LOESS since it requires fewer ( F ^ ST ( m ) , ρ ^ D ( m ) ) points and thus fewer genomic prediction models to be fitted ., In fact , if we assume that the decay rate is linear we could also estimate it as the slope of the line passing through ( F ^ ST ≈ 0 , ρ ^ CV ) and ( F ^ ST ( m ) , ρ ^ D ( m ) ) for a single , small value of m ., It should be noted , however , that several factors can cause departures from linearity , including the number of causal variants underlying the trait , the use of small training populations and the confounding effect of exogenous factors ., In the case of the MICE data , for instance , predictions may be influenced by cage effects; in the case of the WHEAT data , environmental and seasonal effects might not be perfectly captured and removed by the trials’ experimental design ., We also note that the decay curves for traits with small heritabilities will almost never be linear , because ρ ^ D converges asymptotically to zero ., Unlike the results reported in 22 , we do not find a statistically significant difference between the strength of the linear relationship between ρ ^ D and F ^ ST and that between the respective squares ., There may be several reasons for this discrepancy; the simulation study in 22 was markedly different from the analyses presented in this paper , since it used simulated genotypes to generate the population structure typical of dairy cattle and since it used GBLUP as a genomic prediction model ., We also observe that when F ^ ST ( m ) ≈ 0 , both ρ ^ D ( m ) and ρ ^ L are , as expected , similar to the ρ ^ CV obtained by applying cross-validation to the training populations selected from the WHEAT and MICE data ., This suggests that indeed ρ ^ CV is an accurate measure of predictive accuracy only when the target individuals for prediction are drawn from the same population as the training sample , as previously argued by 14 and 19 , among others ., Some limitations of the proposed approach are also apparent from the results presented in the previous section ., The most important of these limitations appears to be that in the context of a breeding program the performance of the decay curve depends on the polygenic nature of the trait being predicted , as we can see by comparing the panels in Fig 1 ., This can be explained by the fact that causal variants underlying less polygenic , highly and moderately heritable traits will necessarily have some individually large effects ., As each of those variants approaches fixation due to selection pressure , allele frequencies in key areas of the genome will depart from those in the training population and the accuracy of any genomic prediction model will rapidly decrease 21 ., However , these selection effects are genomically local and so have little impact on F ^ ST . A similar effect has been observed for flowering time in the WHEAT data ., 4 notes that the Ppd-D1a gene is a major driver of early flowering , but it is nearly monomorphic in one allele in French wheat varieties and nearly monomorphic in the other allele in Germany and the UK ., As a result , even though the F ^ ST for those countries are as small as 0 . 031 and 0 . 042 , ρ ^ D widely overestimates ρ ^ P in both cases ., A possible solution would be to compute F ^ ST only on the relevant regions of the genome or , if their precise location is unknown , on the relevant chromosomes; or to weight F ^ ST to promote genomic regions of interest ., On the other hand , in the case of more polygenic traits a larger portion of the genome will be in linkage disequilibrium with at least one causal variant , and their effects will be individually small ., Therefore , F ^ ST will increase more quickly in response to selection pressure and changes in predictive accuracy will be smoother , thus allowing ρ ^ D to track them more easily ., Indeed , in the WHEAT data the genomic prediction model for flowering time has a much smaller number of non-zero coefficients ( 28 ) compared to yield ( 91 ) , height ( 286 ) and grain protein content ( 121 ) ., Similarly , in the MICE data the model fitted on F010 to predict weight has only 168 non-zero coefficients while others range from 212 to 1169 non-zero coefficients ., By contrast , all models fitted for predicting weight , which correspond to curves that well approximate other families’ ρ ^ P , have between 1128 and 2288 non-zero coefficients ., The simulation on the HUMAN data suggests different considerations apply to outbred species ., Having some large-effect causal variants does not necessarily result in low quality decay curves; on the contrary , if we assume that the trait is controlled by the same causal variants in the training and target populations it is possible to have a good level of agreement between the ρ ^ D and the ρ ^ P . Intuitively , we expect strong effects to carry well across populations and thus ρ ^ D does not decrease beyond a certain FST ., However , this will mean that the curves will not be linear and ρ ^ L will underestimate ρ ^ P ( see Fig 3 , top left panel ) ., We also note that effect sizes are the same in all the populations , which may make our estimates of predictive accuracy optimistic ., Another important consideration is that since the decay curve is extrapolated from the training population , its precision decreases as FST increases , as can be seen from both simulations and by comparing the WHEAT and MICE data ., Predictions will be poor in practice if the target and the training populations are too genetically distinct; an example are rice subspecies 17 , which have been subject to intensive inbreeding ., The trait to be predicted must have a common genetic basis across training and target populations ., However , the availability of denser genomic data and of larger samples may improve both predictive accuracy and the precision of the decay curve for large FST ., Furthermore , the range of the decay curve in terms of FST depends on the amount of genetic variability present in the training population; the more homogeneous it is , the more unlikely that k-means clustering will be able to split it in two subsets with high F ^ ST ( 0 ) ., One solution is to assume the decay is linear and use ρ ^ L instead of ρ ^ D to estimate ρ ^ P; but as we noted above this is only possible if ρ ^ P ≫ 0 ., If ρ ^ P ≈ 0 , the decay curve estimated with LOESS from ρ ^ D can converge asymptotically to zero as F ^ ST increases; but the linear regression used to estimate ρ ^ L will continue to decrease until ρ ^ L ≪ 0 ., Another possible solution is to try to increase F ^ ST by moving observations between the two subsets , but improvements are marginal at best and there is a risk of inflating ρ ^ D . Even with such limitations , estimating a decay curve for predictive correlation has many possible uses ., In the context of plant and anima | Introduction, Materials and Methods, Results, Discussion | The prediction of phenotypic traits using high-density genomic data has many applications such as the selection of plants and animals of commercial interest; and it is expected to play an increasing role in medical diagnostics ., Statistical models used for this task are usually tested using cross-validation , which implicitly assumes that new individuals ( whose phenotypes we would like to predict ) originate from the same population the genomic prediction model is trained on ., In this paper we propose an approach based on clustering and resampling to investigate the effect of increasing genetic distance between training and target populations when predicting quantitative traits ., This is important for plant and animal genetics , where genomic selection programs rely on the precision of predictions in future rounds of breeding ., Therefore , estimating how quickly predictive accuracy decays is important in deciding which training population to use and how often the model has to be recalibrated ., We find that the correlation between true and predicted values decays approximately linearly with respect to either FST or mean kinship between the training and the target populations ., We illustrate this relationship using simulations and a collection of data sets from mice , wheat and human genetics . | The availability of increasing amounts of genomic data is making the use of statistical models to predict traits of interest a mainstay of many applications in life sciences ., Applications range from medical diagnostics for common and rare diseases to breeding characteristics such as disease resistance in plants and animals of commercial interest ., We explored an implicit assumption of how such prediction models are often assessed: that the individuals whose traits we would like to predict originate from the same population as those that are used to train the models ., This is commonly not the case , especially in the case of plants and animals that are parts of selection programs ., To study this problem we proposed a model-agnostic approach to infer the accuracy of prediction models as a function of two common measures of genetic distance ., Using data from plant , animal and human genetics , we find that accuracy decays approximately linearly in either of those measures ., Quantifying this decay has fundamental applications in all branches of genetics , as it measures how studies generalise to different populations . | biotechnology, population genetics, cereal crops, plant science, mathematics, forecasting, statistics (mathematics), crops, plant genomics, mammalian genomics, population biology, plants, research and analysis methods, curve fitting, grasses, crop science, mathematical functions, mathematical and statistical techniques, plant genetics, wheat, animal genomics, agriculture, genetics, biology and life sciences, physical sciences, genomics, evolutionary biology, plant biotechnology, statistical methods, genomic medicine, organisms, human genetics | null |
journal.pgen.1005673 | 2,015 | Human β-D-3 Exacerbates MDA5 but Suppresses TLR3 Responses to the Viral Molecular Pattern Mimic Polyinosinic:Polycytidylic Acid | HBD3 is a member of the β-defensin multigene family ., The amphipathic , antiparallel β-sheet structure , stabilised by disulfide bonds , via six canonical cysteines is conserved throughout evolution 1 and between family members despite significant sequence diversity 2 ., These powerful cationic antimicrobials directly kill fungi , bacteria and viruses , and recently it has become clear that this gene family has roles in other processes including male fertility , immunomodulation and inflammatory disease 3 ., Defensins are primarily expressed from mucosal surfaces , some exclusively in the reproductive tract and others in skin , intestine and gingival surfaces 4–6 ., Of β-defensin genes , hBD3 is probably the most versatile and studies both in vitro and in vivo demonstrate its ability to chemoattractant immune cells 7; encourage wound healing 8 and modulate innate signalling 9–11 ., HBD3 ( gene name DEFB103 ) and its mouse orthologue ( Defb14 ) are promiscuous ligands with ability to bind the receptors CCR6 , CCR2 , CXCR4 ., In addition , a dominant mutation in the DEFB103 gene in dogs and wolves causes an increase in canine β-defensin 3 ( CBD103 ) peptide level allowing off-target binding to melanocortin receptor 1 ( MC1R ) , which results in black coat colour 7 , 12–15 ., DEFB103 is present on hypervariable clusters of six β-defensin genes and alteration in copy number may influence innate immune responses ., Increased copy number of the cluster is associated with psoriasis 16 , 17 ., Increased defensin peptide level has been reported in serum of psoriasis patients , although the influence of defensins on the pathogenesis of the disease is not understood ., Psoriasis is a T cell-mediated disease predominantly orchestrated by Th-17 cells ., Amplification of the disease process is triggered by an initial phase modulated by an increase in innate immune signalling through pattern recognition receptors ( PRR ) such as toll like receptors ( TLR ) 18 ., Psoriatic patients have an increase in dendritic cells and cationic antimicrobial peptides in the skin 19 ., It has been shown that self and viral nucleic acids trigger an increase in the type I Interferon-α response of plasmacytoid dendritic cells ( pDC ) that are specialised cells for Interferon-α production through TLR920 ., Blocking production of Interferon-α by these cells prevents T cell–dependent development of psoriasis in a xenograft model ., The antimicrobial peptide LL-37 has been identified as a molecule that encourages recognition of self DNA and RNA through TLRs on pDC to induce release of Type I Interferon 21 ., Recently antimicrobial peptides hBD2 , hBD3 and lysozyme have also been shown to bind self-DNA and activate pDC through TLR9 to release Interferon-α ., The presence of these peptides in psoriatic plaques suggests a concerted role for them in the pathogenesis of psoriasis 9 , 19 ., Here we determine the effect of hBD3 on the response of primary macrophages to known pathogen-associated molecular patterns ( PAMPs ) to investigate the influence of hBD3 on signalling from innate immune receptors ., We have previously shown that hBD3 suppresses the TLR4-mediated response to bacterial lipopolysaccharide ( LPS ) which is mediated by both MyD88 ( Myeloid Differentiation Primary Response 88 ) and TICAM1 ( Toll-Like Receptor Adaptor Molecule 1-also known as TRIF ) 10 ., We show that polyI:C in the presence of hBD3 has an exacerbated Interferon-β response and decreases CXCL10 production , in vitro and in vivo in both mice and human primary cells ., PolyI:C is a synthetic double stranded RNA ( dsRNA ) and consequently a viral ligand mimic , which is recognised by endosomally located TLR3 and also by cytoplasmic RIG-I-like receptors ( RLRs ) 22 ., Recently high molecular weight ( HMW ) or long poly:IC has been shown to preferentially access RLRs in conventional dendritic cells 23 ., The RLRs include RIG-I ( also known as DDX58 ( DEAD ( Asp-Glu-Ala-Asp ) box polypeptide 58 ) and MDA5 ( Melanoma Differentiation-Associated protein 5 , also known as IFIH1 -interferon induced with helicase C domain 1 ) 24 ., HMW polyI:C is recognised primarily by MDA5 ( without the need for transfection ) and secondarily ( with transfection ) by RIGI 23 ., Activation of MDA5 and consequent interferon-β production has been shown to be associated with autoimmune disorders 25–28 ., Both TLR3 and RLR types of receptor are MYD88 independent , with MDA5 and RIGI requiring the recruitment of the adaptor protein MAVS ( mitochondrial antiviral signalling protein -also known as VISA , IPS-1 and CARDIF ) ., Signalling through the MAVS pathway in macrophages results in IRF3/7 driven expression of type I Interferon and NF-ᴋB induction of inflammatory cytokines and chemokines 29 , 30 ., TLR3 specifically associates with the adaptor TICAM1 and mediates signal transduction which activates IRF3 and NF-ᴋB 30 ., Our studies have been carried out using HMW polyI:C throughout ., We dissect the effect on the pathways responsible for the altered response of macrophages to HMW polyI:C in the presence of hBD3 and reveal the mechanism that enables increased Interferon-β and decreased CXCL10 production ., We previously reported that hBD3 suppresses Toll Like Receptor 4 ( TLR4 ) -induced signalling in response to LPS both through MYD88 and TICAM1 pathways 10 ., To investigate the effect of hDB3 on other TLR pathways we exposed primary bone marrow derived macrophages from mice ( BMDM ) to a variety of TLR ligands and found that the response to TLR 2 , 2/6 , 1/2 , 7 or 9 ligands were not significantly altered in the presence of hBD3 ( Fig 1A ) ., In agreement with our previous findings , hBD3 inhibited LPS-induced TNFα ., In contrast , the response to HMW polyI:C was significantly exacerbated in the presence of hBD3 , increasing TNF-α; Interferon-β and IL-6 production , however hBD3 significantly suppressed the CXCL10 ( also known as IP10 ) response to polyI:C ( Fig 1B ) ., The enhancing effect of hBD3 on polyI:C cytokine induction , was also seen in the human monocytic cell line THP1 , where hBD3 significantly enhanced polyI:C-induced TNFα , IL-6 and IL-8 ., However , in contrast to what we observe in mouse cells , there was no significant effect of hBD3 on polyI:C-induced CXCL10 in human cells ( Fig 1C ) ., In addition this enhancing effect was also seen on Interferon-β gene expression in primary human peripheral blood monocyte derived macrophages ( PBMDM ) , measured by qRT-PCR ( Fig 1D ) ., The amount of hBD3 required to induce an enhanced Interferon-β response to polyI:C differed from that required to decreased CXCL10 response ( Fig 2A ) ., PolyI:C-induced CXCL10 was inhibited by 4μg/ml hBD3 , but no longer inhibited at 2μg/ml hBD3 , whereas concentrations as low as 0 . 05 μg/ml hBD3 enhanced the Interferon-β response to polyI:C , although at this concentration the effect is beginning to diminish ., We tested the importance of the hBD3 cysteine-stabilised structure using hBD3 with the canonical defensin motif of six cysteines ( which form three intramolecular disulfide bonds ) replaced with serines ., This modified peptide ( hBD3 Cys:Ser ) did not augment cytokine production in response to polyI:C , suggesting that the enhancing effect of hBD3 on polyI:C signalling is dependent on the 3-dimensional structure of the hBD3 peptide ( Fig 2B ) ., This lack of effect was not due to the inability of the linear peptide to rapidly enter the cell as TAMRA ( tetramethylrhodamine azide ) -labelled hBD3 Cys:Ser peptide entered BMDM in 10 min ( Fig 2B ) similar to canonically folded hBD3 , as shown in our previous studies10 ., To test whether these in vitro effects were relevant in vivo , we injected wild type mice with polyI:C in the presence of synthetic hBD3 peptide ( Fig 3A ) ., We found that TNF-α and IL12p40 responses to polyI:C were significantly increased in the presence of synthetic hBD3 ., Interestingly the cytokine CXCL10 was not decreased in the presence of hBD3 ( Fig 3A ) although there was a trend for CXCL10 to be lower in the presence of hBD3- in keeping with the significant reduction of this cytokine seen in the BMDM stimulated with polyI:C and hBD3 ., There is a concern that unless synthetic hBD3 is correctly oxidised and the correct disulphide bonding achieved , the properties of the peptide may be altered 7 ., The importance of structure is confirmed in our experiment with hBD3 Cys:Ser shown in Fig 2B ., The oxidised synthetic peptide we used here ( obtained from Peptide Institute , Japan ) gives details of preparation and oxidation 31 and implies correct cysteine bonding ( C1-C5; C2-4 and C3-C6 ) ., However , in order to investigate our effect with hBD3 oxidised in vivo , we expressed the gene ( DEFB103 ) that encodes hBD3 , as a transgene in mice ., Transgenic mice were made using the pCAAG promoter ( as described previously by Candille et al . 14 ) by introducing a transgene containing the genomic copy of DEFB103 into ES cells ., The vector ( shown in S1A Fig ) expresses hBD3 only after CRE recombinase-mediated deletion of the floxed DsRed , puromycin STOP spacer ., After deletion , hBD3 and EGFP are expressed as a polycistronic mRNA and translated as independent proteins using an IRES site ., We made several ES cell lines with the DEFB103 expression construct and isolated a clone with strong expression by virtue of DsRed expression and a single site of insertion of the transgene ( located to the sub-telomeric region of chromosome 12 by FISH and chromosome painting ( S1B Fig ) ., The control DsRed transgenic mice were made with this ES clone ., HBD3 transgenic mice were made from the same ES clone after treatment with CRE recombinase ., These cells had strong expression of EGFP and hBD3 and only the ES cells containing the CRE-excised vector showed expression of DEFB103 mRNA ( S1C Fig ) ., HBD3 mRNA expression was detected in all tissues tested from mice made using the hBD3-Tg ES cells and hBD3 protein was detected in various tissues including BMDM by immunohistochemistry ( S1D Fig ) using an hBD3 monoclonal antibody 32 ., No hBD3 was detected in DsRed-Tg control mice ., The effect of polyI:C exposure in mice expressing physiologically secreted hBD3 was tested by exposing heterozygote hBD3-Tg and DsRed-Tg mice to polyI:C ( 100μg , i . p . ) ., The transgenic animals expressing hBD3 demonstrated an increased level of both Interferon-β and TNF-α ( Fig 3B ) ., In addition , homozygous transgenic hBD3 mice had a significantly raised basal level of Interferon-β compared to controls ( Fig 3C ) ., A number of additional cytokines were investigated , including IL-6 and CXCL10 ., hBD3-Tg mice treated with polyI:C showed a trend towards enhanced IL-6 induction and reduced CXCL10 induction compared to controls , however these data did not reach significance ( see S2 Fig ) ., HMW polyI:C enters macrophages without a transfection reagent , however complexing with the cationic lipid , Lipofectamine 2000 , enhances the amount of polyI:C entering cells by endocytosis , allowing more ligand to be available to both endosomal and cytoplasmic receptors 33 ., We hypothesised that hBD3 , being positively charged , may form complexes with polyI:C , enhancing uptake in a similar way to lipofection ., To investigate this , we compared cellular uptake of FITC-labelled polyI:C in the presence of Lipofectamine 2000 and/or hBD3 , by flow cytometry ., Compared to polyI:C alone , the addition of hBD3 significantly increased the amount of labelled ligand per cell ( Fig 4A ) ., Lipofectamine alone also significantly augmented the amount of polyI:C entering cells , with the amount of FITC-polyI:C uptake in the presence of Lipofectamine not differing significantly from the level of FITC-polyI:C in hBD3 treated cells ., Uptake in the presence of both lipofectamine and hBD3 was similar to either hBD3 or lipofectamine alone , so no additive effect was apparent ., We further compared the effects of lipofectamine and hBD3 on the production of Interferon-β and CXCL10 by polyI:C ., Transfecting polyI:C with lipofectamine into wildtype BMDM resulted in enhanced CXCL10 production ( Fig 4B ) ., In contrast hBD3 inhibited CXCL10 induction by polyI:C ., In the presence of a combination of lipofectamine and hBD3 , CXCL10 production was still inhibited compared to polyI:C and lipofectamine , but to a lesser extent than hBD3 alone ., In contrast , hBD3 and hBD3/lipofectamine significantly increased Interferon-β production in response to polyI:C in BMDM ( Fig 4C ) ., Lipofectamine and polyI:C did not demonstrate an enhanced Interferon-β response and lipofectamine decreased the amount of Interferon-β produced in response to polyI:C ., To visualise the uptake of polyI:C by BMDM we used FITC-labelled polyI:C ( green fluorescence ) ., Stronger fluorescence , was observed in cells when either hBD3 or lipofectamine was also present , supporting the flow cytometry data in Fig 4A showing both hBD3 and lipofectamine increase the entry of polyI:C into the cells ., HBD3 increased the intensity of cyloplasmic polyI:C staining compared to either lipofectamine or polyI:C alone ., In the presence of lipofectamine , polyI:C had a more punctate pattern within the cell , consistent with endosomal location ., In the presence of hBD3 , polyI:C appeared to have increased cytoplasmic distribution in addition to foci of staining ., These results show that lipofectamine and hBD3 , both enhance the amount of polyI:C that enters the cell but the different signalling responses triggered by each , suggests that hBD3 is directing polyI:C towards cellular compartments , that are not those targeted by the polyI:C-lipofectamine complexes ., To further investigate the mechanism of the hBD3 effect on polyI:C we carried out fluorescence co-localisation studies using TAMRA-labelled hBD3 , FITC-labelled polyI:C and immunohistochemistry for the early endosomal marker ( EEA1 ) ., TAMRA-labelled hBD3 entered the cells by 10mins ( Fig 5Aii ) , whereas FITC-polyI:C was not visible in the cells until 30 minutes ( Fig 5Biii ) ., FITC-polyI:C added to the cells without hBD3 ( Fig 5B and 5C ) , demonstrated FITC fluorescence in punctate regions which stained to some extent with the early endosome marker EEA1-1 antibody ( Pearson coefficient , r = 0 . 615; Fig 5Civ ) ., HBD3 alone ( Fig 5A and 5D ) also localised to discrete regions however these did not co-localise with EEA1 staining ( Pearson coefficient , r = 0 . 205; Fig 5Div ) ., Adding TAMRA-hBD3 and FITC-polyI:C together onto BMDM , again showed more polyI:C entering the cell in the presence of hBD3 ( Fig 5Biii , 5Cii vs 5Eiii ) ., However we could see no evidence for co-localisation of hBD3 and polyI:C ( Pearson coefficient , r = 0 . 073; Fig 5Evii ) ., PolyI:C appeared more cytoplasmic in the presence of hBD3 than when it entered the cell alone and the co-localisation coefficient with EEA1-1 staining was lower than in the absence of hBD3 ., ( Pearson coefficient , r = 0 . 562 compared to r = 0 . 615 Fig 5Civ versus Fig 5Evi ) ., In the presence of polyI:C , hBD3 remained localised in the discrete foci similar to those seen with hBD3 treatment alone , and again hBD3 did not co-localise with EEA1 ( Pearson coefficient r = 0 . 05; Fig 5Eviii ) ., This implies that hBD3 , alters the localisation of polyI:C allowing less polyI:C to access the early endosome ., Although nucleic acids can induce type I Interferon by activation of TLR signalling 34 in the endosome , Interferon-β and IL-6 can also be produced by activation of cytoplasmic receptors ., To examine the consequences of the altered localisation of polyI:C by hBD3 and to determine the signalling pathways responsible for production of Interferon-β and CXCL10 , we used cells from knockout mice specific for the two main pathways known to be involved ., Firstly , we determined the Interferon-β response of BMDM exposed to polyI:C in the absence of the TLR3 adaptor TICAM1 and found that this was reduced in Ticam1-/- cells compared to wild type BMDM , indicating that the majority of the polyI:C effect on Interferon-β was not TICAM1 dependent ., In the presence of hBD3 , the polyI:C-induced Interferon-β response in Ticam1-/- BMDM was still significantly enhanced compared to polyI:C alone ( Fig 6A ) , which suggests that hBD3 is not enhancing signalling through the TLR3/TICAM1 pathway ., Conversely , in the absence of MAVS , where Interferon production is only through TLR3/TICAM1 signalling , a smaller amount of Interferon-β was induced by polyI:C ., In the presence of hBD3 this Interferon-β induction was significantly inhibited ( Fig 6A ) , suggesting that hBD3 inhibits TLR3/TICAM1 signalling ., Treatment of BMDM from Ticam1 ( -/- ) Mavs ( -/- ) double knockouts with polyI:C did not induce Interferon-β indicating that all the Interferon-β response to polyI:C in BMDM is dependent only on these two pathways ( S4A Fig ) ., In contrast to Interferon-β , the induction of CXCL10 by polyI:C in wildtype BMDM was significantly reduced in the presence of hBD3 ., In Ticam1 ( -/- ) BMDM the response to polyI:C was eliminated ( Fig 6B ) demonstrating that production of this cytokine is controlled primarily by TLR3/TICAM1 activation ., Supporting this finding , polyI:C-induced CXCL10 in Mavs ( -/- ) BMDM ( a TLR3-TICAM1 dependent response ) was not significantly different to wildtype cells indicating that MAVS does not influence CXCL10 production and the production of CXCL10 in response to polyI:C in Mavs ( -/- ) cells was still significantly inhibited by hBD3 ( Fig 6B ) ., Both RIGI and MDA5 are upstream of MAVS , so to dissect the effects of hBD3 on MAVS signalling we used the specific RIGI ligand , 5’ triphosphate double stranded RNA ( 5’ppp ) and Mda5 ( -/- ) mice 35 ., Treatment of wildtype BMDM with 5’ppp resulted in a significant increase in Interferon-β , TNFα and CXCL10 , and as expected with this ligand , these responses were dependent on the presence of MAVS ( Fig 7A ) ., The addition of hBD3 to the macrophages shortly after transfection of 5’ppp , resulted in a significant decrease in cytokine and Interferon-β production ( Fig 7A ) , demonstrating that RIGI responses to 5’ppp are inhibited by the presence of hBD3 ., In contrast however , macrophages from Mda5 ( -/- ) mice , revealed that Interferon-β induced by polyI:C was reduced compared to wildtype cells , indicating that MDA5 signalling was responsible for the majority of the polyI:C induced Interferon-β production ( Fig 7B ) ., This residual response which is likely to be TLR3-TICAM1 signalling , was not amplified in the presence of hBD3 , indicating that MDA5 is required for the hBD3 enhancing effects on polyI:C-induced Interferon-β ., We show here that hBD3 enhances the production of various cytokines in response to polyI:C ( TNF-α , IL6 in mouse cells and TNF-α , IL8 in human cells ) and Interferon-β in both human and mouse cells ., We demonstrate that this effect is dependent on the correct , disulfide stabilised structure of hBD3 ., Importantly , we show that this response is not specific to our synthetic hBD3 peptide , as transgenic animals expressing hBD3 from a genomic transgene , also demonstrate an increased type I Interferon response when injected with polyI:C ., It was important to reproduce the results shown with our synthetic hBD3 peptide in this transgenic system to validate the augmenting effects of hBD3 as it has been demonstrated that synthetic peptides may be incorrectly folded giving misleading results 7 ., The synthetic hBD3 peptide we use here , gives equivalent functional results to hBD3 produced in vivo ., When we dissect the pathways known to be activated by polyI:C , it is evident that although signalling through the RLR co-adaptor MAVS mediated pathway is up-regulated in the presence of hBD3 , signalling through the endosomally located TLR3/TICAM1 pathway is suppressed ., The inhibition of the TICAM1-mediated signalling pathway supports our findings with LPS , where we previously reported that hBD3-mediated inhibition of LPS signalling through TLR3/TICAM1 was lost in Ticam1 KO mice and could be inhibited by hBD3 cDNA in HEK293 cells 10 ., In our experiments here , we use HMW ( long ) polyI:C which activates MDA5 with or without transfection and show that in BMDM , MDA5 is predominantly responsible for the Interferon-β response ., Interestingly , it has been demonstrated previously that LMW and HMW polyI:C are recognised by different receptors with HMW polyI:C being recognised by MDA5 and LMW polyI:C by RIGI 24 ., In addition large RNA structures generated by viral replication are believed to be important in effectively triggering MDA5 36 ., Recently Zou et al reported that forced delivery of HMW polyI:C to the cytoplasm with transfection was not necessary for RLR stimulation in GM-DCs and CD11bhiCD24lo DCs , although LMW polyI:C required transfection to interact with MDA5 23 ., We show here that similarly to the DCs , BMDM also take up HMW polyI:C effectively without transfection and activate MDA5 ., HBD3 exacerbates the signalling through MDA5 ., These researchers also report release of endosomal Cathepsin D and induction of necrosis by the activation of MDA5 ., We see no evidence of cell death ( using an LDH assay , see S5 Fig ) using polyI:C at 10μg/ml , but this is 5-fold less than that used by Zou et al 23 ., Cationic lipids such as Lipofectamine are known to cause endosomal localisation 33 ., Although hBD3 is a cationic peptide we do not observe similar outcomes when we compare the effects of hBD3 with the actions of lipofectamine on polyI:C stimulation of macrophages ., In wild type cells stimulated with polyI:C , hBD3 increased Interferon-β and decreased CXCL10 production ., In contrast , polyI:C and lipofectamine increased CXCL10 production and decreased Interferon-β ., This effect is likely to be due to the change in localisation of polyI:C as a result of being in the presence of lipofectamine or hBD3 ( see Fig 8 ) ., Our immunostaining demonstrates that hBD3 encouraged polyI:C to be more cytoplasmic compared to lipofectamine which causes increased endosomal localisation ., Despite the likely electrostatic interaction of polyI:C and the highly charged hBD3 ( +11 ) in the cell , our cellular uptake experiments using fluorescently labelled derivatives revealed that at 30min after addition to the cells hBD3 and polyI:C do not co-localise appreciably ., It is possible that initially they may have interacted , allowing polyI:C to access the cell as hBD3 has been described as having cell penetrating properties 37 ., In the presence of hBD3 , polyI:C does not localise to the early endosome so presumably the cytoplasmic location of the ligand allows an increase in interaction with MDA5 ., It is possible that cationic hBD3 complexed with polyI:C , enables the ligand to rapidly escape the acidic endolysosome , perhaps in a similar way to pH-dependent fusogenic peptides that assist macromolecules to access the cytoplasm 33 ., However we see polyI:C localised to the early endosome ( by EEA1 positive immunostaining ) in the presence of hBD3 implying that the structure of the early endosome is not disrupted by the presence of hBD3 ., The main consequence of increased MDA5 signalling in response to polyI:C is increased IFN-β ., This increase is additive when lipofectamine is also present ( S4B Fig ) ., However no increase is observed in the absence of MAVS which implies that the exacerbated response requires the RLR ., It may be that lipofectamine complexes the polyI:C to create higher order structures that activate MDA5 more optimally when 36 hBD3 increases its cytoplasmic localisation ., MDA5 is important in relation to autoimmunity and mutations that inactivate or reduce expression of MDA5 have been shown to protect individuals from type I diabetes mellitus risk 38 , 39 ., In addition , a mutant form of MDA5 in mice that is active without viral infection induces a type I Interferon-dependent autoimmunity with similarities to lupus 25 ., However Interferon-β has also been described as a protector against some types of inflammation such as dextran sodium sulphate induced colitis 40–42 and this protection can be observed in mice that express increased Interferon-β in response to dsRNA-producing intestinal bacteria 40–42 ., Increased copy number of the cluster of β-defensins on human chromosome 8 is linked to increased incidence of the autoimmune disease psoriasis and effective treatment of psoriasis with UV irradiation is linked to suppression of type I Interferon and Th17 cells 43 ., In addition the most effective treatments currently for psoriasis are monoclonal antibodies directed against IL-17 cytokine production or IL-12p40 ( the cytokine subunit common to both IL-12 and IL-23 ) 18 , 44 ., It is thus potentially highly significant that we see strong elevation of IL12-p40 subunit in mice injected with both hBD3 and polyI:C ., It is also possible that the other defensins on the CNV cluster may also demonstrate this effect and we have shown that hBD2 also heightens the response of mouse BMDM to polyI:C ( S6 Fig ) ., It has recently been shown that human pDC produce Interferon-α in response to self or other DNA through TLR9 9 , 19 ., We show here that macrophages increase the Interferon-β response to polyI:C in the presence of hBD3 through MDA5 ., Production of type I Interferons is normally the consequence of pattern recognition receptors binding virally produced nucleic acid pathogen associated molecular patterns ( PAMP ) , such as double stranded RNA ( dsRNA ) produced during viral replication ., Psoriasis has been reported to be exacerbated by the use of Interferon-α as therapy for Hepatitis C 45 and by Interferon-β therapy for multiple sclerosis 46 ., Investigation of the psoriasis transcriptome has identified an increase in RIG-I like receptors ( RLR ) , which also recognise viral PAMP leading to type 1 Interferon production 47 ., During a pathogen infection , hBD3 expression increases 32 , 48 , and hBD3 has been shown to demonstrate potent anti-viral action in vitro 49 ., Expression in pDC , monocytes and epithelial cells of the non-copy number variable defensin hBD1 has been shown to increase in response to virus exposure , while expression of the murine orthologue of DEFB103 ( Defb14 ) increases in response to polyI:C 50 , 51 ., MDA5 is specialised for protecting mice against infection with various RNA viruses including picornaviruses ( including Theiler’s and Mengo viruses and Encephalomyocarditis virus ( EMCV ) ) as well as paramyxovirus and Norovirus 52 , 53 ., MDA5 knockout mice are highly susceptible to EMCV 35 , 54 ., During infection , rapid killing , detection and innate response are essential; therefore in this regard , high hBD3 copy number and potentiation of PRR may be beneficial ., However an undesirable effect of increased copy number of the defensin cluster ( and concomitant increase in expression of defensin peptides ) may be over stimulation of PRRs leading to exuberant production of type I interferons ., This double edged sword may provide protection against pathogens in the short term , but in the longer term contribute to the development of psoriasis in individuals with an increased copy number of the β-defensin cluster ., Animal studies were covered by Project License ( PPL 60/4475 ) , granted by the UK Home Office under the Animal Scientific Procedures Act 1986 , and locally approved by the University of Edinburgh Ethical Review Committee ., Human venous blood was collected with written patient consent from healthy volunteers according to Lothian Research Ethics Committee approvals ♯08/S1103/38 ., Ultra pure Lipopolysaccharide ( LPS ) from E . coli 0111:B4 , Lipoteichoic acid ( LTA ) , Pam3CSK4 , FSL-1 , HKLM , polyI:C ( HMW ) , FlTC-labelled polyI:C ( HMW ) , R848 , CpG and 5’ triphosphate double stranded RNA ( 5’ ppp-dsRNA ) were purchased from InvivoGen ( San Diego , USA ) , M-CSF , ELISA DuoSets and IFNβ antibodies were obtained from R&D Systems , ( Abington , UK ) ., Fluorescently labelled secondary antibodies were purchased from Jackson ImmunoResearch Laboratories ( PA , USA ) ., hBD3 ( GIINTLQKYYCRVRGGRCAVLSCLPKEEQIGKCSTRGRKCCRRKK ) was from Peptides International , and cys-ser hBD3 and cys-ser-TAMRA hBD3 were from Almac ( Almac Group Ltd , Craigavon , UK ) ., The peptide was produced on a CEM Liberty1 microwave peptide synthesizer using standard Fmoc ( fluorenylmethyloxcarbonyl chloride ) chemistry ., Amino acids were purchased from AAPPTec and were assembled on H-Rink amide ChemMatrix resin ., Fmoc protecting groups were removed using 20% piperidine and 0 . 1 M hydroxybenzotriazole ( HOBt ) in dimethylformamide ( DMF ) ., Amino acids were coupled using 5 molar equivalents of diisopropylcarbodiimide ( DIC ) and 10 molar equivalents of HOBt in DMF ., The N-terminal labelling of peptides with fluorescent dye was performed on resin-bound peptide using 4 equivalents of 5- ( and-6 ) -carboxytetramethylrhodamine , succinimidyl ester ( 5 ( 6 ) -TAMRA SE purchased from Biotium ) and 6 equivalents of diisopropylethylamine ( DIEA ) in DMF , incubating for 2 hours ., The peptide resin was then rinsed with DMF to remove excess fluorescent dye , washed with dichloromethane ( DCM ) , and dried ., Cleavage of the peptide from resin was performed in a trifluoroacetic acid ( TFA ) / triisopropylsilane ( TIS ) / 1 , 2-ethanedithiol ( EDT ) / phenol ( 90:4:4:2 ) mixture for 90 min ., The resin was filtered and the filtrate was added to 90 mL of cold dry diethyl ether ., The precipitate was collected by centrifugation and the diethyl ether was discarded ., The peptide was purified on a C18 reverse phase HPLC column and the correct molecular weight was confirmed by ESI-MS ., Oxidative folding was achieved in folding buffer ( 0 . 5–1 . 0 M guanidine hydrochloride ( GuHCl ) , 0 . 1 M Tris , 1 mM glutathione ( GSH ) , 0 . 1 mM oxidized glutathione ( GSSG ) , pH 8 . 5 ) at a peptide concentration of 0 . 1 mg/mL and stirred for 48 hours ., Folding was monitored by reverse phase HPLC , which revealed one major species that was used in subsequent experiments ., Folding procedures were developed to give the correct HBD3 structure , as verified previously by nuclear magnetic resonance structure determination ( Nix et al . , 2013 ) ., The folded products were purified on a C18 reverse phase HPLC column and identified as fully oxidized peptides by ESI-MS ., Quantitative concentrations were determined with amino acid analysis at the molecular structure facility at UC Davis ., The Mavs−/− ( Cardif , Ips-1 , or Visa ) mutant line was generated by the Tschopp group in Lausanne ., It is homozygous-viable null mutant in the C57B6J background ., Ticam1 ( Trif ) -/- mice and MDA5-/- mice were used with the generous permission of Professor Shizuo Akira ( Osaka University , Japan ) 35 , 55 ., hBD3-Tg and DsRed-Tg were constructed by electroporation of ScaI linearised parental vector , which has a 1 . 5 Kb genomic fragment of the entire hBD3 gene DEFB103 including exons 1 and 2 and the intervening intron cloned into pTLC plasmid ( a kind gift from Josh Brinkman , Danish Stem Cell Centre , DanStem , University of Copenhagen ) using DEFB103 primers with 5 NheI and PacI sites for cloning , into ES cells ., Cells that strongly expressed DsRed were selected by FACS and used to make DsRed-Tg mice ., Transient cre treatment of these cells produced DsRed negative cells due to lox -mediated excision of the DsRed gene and allowed expression of hBD3 and EGFP ( see S1 Fig ) ., The ES cells were made into macrophages using the method of ( Yeung et al 2015 ) which were strong expressors of EGFP ., Clones before and after CRE treatment were injected into blastocysts at the University of Edinburgh MRC Evans Building , Transgenic Unit ., THP-1 cells were grown in RPMI with 10% fetal bovine serum ( FBS ) and differentiated into macrophages by the addition of 150nM PMA , 2 days before treatment ., Mouse primary macrophages ( BMDM ) were generated from femur bone marrow grown for 8 days in DMEM with 10% fetal bovine serum and 20ng/ml M-CSF ., Cells , seeded at 2 x 105 in | Introduction, Results, Discussion, Materials and Methods | Human β-defensin 3 ( hBD3 ) is a cationic host defence peptide and is part of the innate immune response ., HBD3 is present on a highly copy number variable block of six β-defensin genes , and increased copy number is associated with the autoimmune disease psoriasis ., It is not known how this increase influences disease development , but psoriasis is a T cell-mediated disease and activation of the innate immune system is required for the initial trigger that leads to the amplification stage ., We investigated the effect of hBD3 on the response of primary macrophages to various TLR agonists ., HBD3 exacerbated the production of type I Interferon-β in response to the viral ligand mimic polyinosinic:polycytidylic acid ( polyI:C ) in both human and mouse primary cells , although production of the chemokine CXCL10 was suppressed ., Compared to polyI:C alone , mice injected with both hBD3 peptide and polyI:C also showed an enhanced increase in Interferon-β ., Mice expressing a transgene encoding hBD3 had elevated basal levels of Interferon-β , and challenge with polyI:C further increased this response ., HBD3 peptide increased uptake of polyI:C by macrophages , however the cellular response and localisation of polyI:C in cells treated contemporaneously with hBD3 or cationic liposome differed ., Immunohistochemistry showed that hBD3 and polyI:C do not co-localise , but in the presence of hBD3 less polyI:C localises to the early endosome ., Using bone marrow derived macrophages from knockout mice we demonstrate that hBD3 suppresses the polyI:C-induced TLR3 response mediated by TICAM1 ( TRIF ) , while exacerbating the cytoplasmic response through MDA5 ( IFIH1 ) and MAVS ( IPS1/CARDIF ) ., Thus , hBD3 , a highly copy number variable gene in human , influences cellular responses to the viral mimic polyI:C implying that copy number may have a significant phenotypic effect on the response to viral infection and development of autoimmunity in humans . | Defensins are classically known as antimicrobial peptides due to their ability to rapidly kill pathogens including bacteria , viruses and fungi ., They are produced in the presence of infectious agents at body surfaces exposed to the environment ., Increasingly , their functional repertoire is expanding , and they have been shown to modulate the immune system ., In humans , there is a block of six β-defensin genes that varies in copy number in the population ., Individuals with an increased number of β-defensin genes have an increased likelihood of developing the skin autoimmune disease psoriasis ., It is not known how this increase in gene copy number influences development of the disease , and psoriasis is a complex interplay of genomic and environmental factors that trigger disease progression and include exposure to viruses ., We examined whether a molecular pattern characteristic of viruses produces an altered immune response in the presence of the defensin human β-defensin 3 ( hBD3 ) ., We find that hBD3 triggers a larger interferon defence response to this viral mimic by increasing accessibility to a cellular receptor that recognises viral patterns ., Interferon is known to be important in autoimmunity and our work may explain why individuals with increased β-defensin number are predisposed to develop psoriasis . | null | null |
journal.pntd.0003645 | 2,015 | Molecular Epidemiology of Rabies Viruses Circulating in Two Rabies Endemic Provinces of Laos, 2011–2012: Regional Diversity in Southeast Asia | Globally , an estimated 60 , 000 people die of rabies annually , and more than 31 , 000 of these deaths occur in Asia 1 , 2 ., Among various Asian regions , the countries of the Association of South East Asian Nations ( ASEAN ) have been working towards substantial economic development ., A call for elimination of rabies by 2020 highlights the political importance of rabies control 3 ., Seven out of the 10 ASEAN member states are rabies-endemic: Cambodia , Indonesia , Laos , Myanmar , the Philippines , Thailand and Vietnam 3 ., Among these countries , Laos has the lowest reported number of human deaths from rabies ., However , the precise morbidity , mortality and molecular epidemiology of rabies in Laos is largely unknown ., This is due to difficulty in collecting data and samples from remote areas of the country , and to the country’s modest data collection system at the centralized diagnostic facility ., Survey of prevalence of canine rabies is still in its initial stages and limited to a small central part of Laos ., The first reported phylogenetic study showed that Laotian rabies viruses are grouped with viruses from Thailand , Myanmar , Cambodia and Vietnam 4 ., Combined analyses from geographic information system data showed that China is the likely source of the Asian rabies viruses , whereas individual migration event suggested that Cambodia may be a source of Asian rabies viruses transmission to China , Laos and Thailand 5 ., The Laotian government vision is to convert the country from a ‘land-locked’ to ‘land-linked’ due to its strategic position in the ASEAN region ., As a result , a number of bridges and new roads have been built to connect Laos to neighboring countries ., Since ancient times , Laos has relied on its neighbors for trading ., For historical , travel convenience and cultural reasons , Thailand has been a primary trading partner ., The Mekong River region is well known for its culture of dog-meat consumption 6 ., With growing economic prosperity and improved roads across this region , movement of people and allegedly , dogs for meat consumption has been increasing substantially ., These factors may have influenced the dissemination and evolution of rabies viruses ., That rabies is a neglected tropical disease is exemplified by the fact that with the exception of Thailand , a full viral genome sequence has been unavailable across the rabies-endemic South East Asian countries ., Rabies belongs to the genus Lyssavirus of the family Rhabdoviridae ., It is a single-stranded , negative-sense RNA of approximately 12 kb that encodes five structural proteins: nucleoprotein ( N ) , phosphoprotein ( P ) , matric protein ( M ) , glycoprotein ( G ) , and RNA-dependent RNA polymerase ( L ) ., These genes are separated by intergenic regions of variable length ., Phylogenetic classification of rabies viruses is based on specific genes , mostly from partial gene sequences , and thus may lead to inconsistent results ., It is possible that phylogenetic classification using complete genome sequences , rather than from a partial sequence of a specific gene , would offer a more robust and comprehensive means of addressing the evolution , spread and genome-wide heterogeneity of viruses 7 ., Complete genome sequencing of rabies viruses from these countries will also help explain the evolution of the viruses in Laos and the Mekong River region ., Because only complete genome sequence can reveal the length , nucleotide substitution patterns , and variations in the start signal , stop signal and other motifs across the genes ., As a result it can provide more information to compare and can generate reliable results on genome evolution ., Therefore , this study was undertaken to quantify the current animal rabies occurrence in Laos and to complete a molecular characterization of the viruses in current circulation ., The study was approved ( No . 38/NECHR ) by the National Ethics Committee for Health Research , Ministry of Health , Laos ., To determine the prevalence of rabies , we analyzed data from samples submitted from various regions of Laos to the rabies laboratory of the National Animal Health Centre , Department of Fisheries and Livestock , Ministry of Agriculture and Forestry from January 2004 through December 2011 ., This is the only central animal rabies diagnostic laboratory in Laos ., In Laos , dog and cat rabies are notifiable , but not wildlife rabies ., For animal samples , the head of the suspected animal was submitted to the laboratory , and the brain was dissected by a trained technician ., These animals had signs of rabies such as aggressive behavior , a tendency to bite and excessive salivation , with or without a history of biting humans and/or animals ., There are no community veterinarians in Laos; samples from suspect animals are submitted mainly by the general public ., Brain sample tests are performed using a fluorescent antibody test ( FAT ) on crushed smears of hippocampus for a fee of 30 , 000 Kip ( approximately US$ 3 . 70 ) and test results are provided on paper to the individual who submitted the sample ., In emergency cases , such as when postexposure prophylaxis ( PEP ) is required , the person is informed of the results by phone ., The samples submitted from Vientiane capital , Vientiane and Champasak provinces ( Fig 1 ) to the rabies laboratory of National Animal Health Centre , during February 2011 through March 2012 , were used in this study for molecular characterization ( S1 Table ) ., Total RNA was extracted from about 1 g of brain homogenate by using Trizol ( Invitrogen , Carlsbad , CA , USA ) according to the manufacturer’s instruction ., Extracted RNA was stored at -30°C until further analyses ., Using random hexamer primers cDNA was synthesized by SuperScript III First-Strand Synthesis System ( Invitrogen ) according to the instructions of the manufacturer ., The synthesized cDNA was diluted with DNase/RNase free water ( Invitrogen ) as a template ., PCR was performed using TaKaRa ExTaq ( Takara Bio Inc . , Shiga , Japan ) ., Nested PCR for nucleoprotein ( N ) gene was performed to detect the presence of rabies virus genome in the samples 8 ., The PCR for N and glycoprotein ( G ) genes , and G-L intergenic region was performed 7 ., Whole genome sequence of two strains , arbitrarily selected from Vientiane capital and Champasak samples was performed 7 ., For 13 other strains only full-length N gene sequence could be performed ., Cycle sequencing of the amplified product was performed using the BigDye Terminator v3 . 1 Cycle Sequencing Kit ( Applied Biosystems , Foster city , CA , USA ) ., The purified amplicons were sequenced using ABI-3130 Genetic sequencer ( Applied Biosystems ) ., All steps were done according to the manufacturer’s instruction ., The 5′ or 3′-terminal end of the genome was amplified using SMART RACE cDNA Amplification Kit ( Clontech Laboratories , Inc . , Mountain View , CA , USA ) according to the instructions of the manufacturer ., Evolutionary analysis was done using the full-length N gene ., We inferred a Maximum Clade Credibility phylogenetic tree using the Bayesian Markov Chain Monte Carlo method available in the BEAST package , v1 . 6 . 1 9 ., The analysis utilized a relaxed ( uncorrelated lognormal ) molecular clock and GTR+Γ+I model of nucleotide substitution ., The model was selected on the basis of Akaike Information Criterion using jModelTest software 10 ., All chains were run for 60 million generations and sampled every 3000 steps ., This resulted in an effective sample size of >521 for all estimated parameters ., The posterior densities were calculated using with 10% burn-in and checked for convergence using Tracer , v . 1 . 5 ., The nucleotide and amino acid sequences of genes and intergenic regions were compared among rabies viruses from Laos ., The nucleotide sequences were used to construct the whole genome phylogenetic tree 7 ., Multiple sequence alignment was done by ClustalW ver . 2 then phytogenic tree was constructed with MEGA ver . 5 using the neighbor joining ( NJ ) method and the branching pattern was statistically evaluated by bootstrap analysis of 1000 replicates ., During this eight years period , number of submitted samples peaked in 2008 and gradually decreased ( Table 1 ) ., Whereas number of rabies positive samples varied annually , peaked in 2009 and gradually decreased ., However the percentage of rabies positive samples increased significantly from 40 . 5% in 2004 to 60 . 2% in 2009 and continued at this level ., Annually , mean 157 . 6 {95% confidence interval ( CI ) 129 . 9–185 . 3} samples were submitted for rabies diagnosis , among them 80 . 6 ( 95% CI 65 . 3–95 . 9 ) samples were rabies positive i . e . 51 . 7% ( 95% CI 45 . 0–58 . 4% ) of the submitted samples were rabies positive ., Most of the submitted samples were from dogs 155 . 1 ( 95%CI 129 . 0–183 . 2 ) , followed by cats 1 . 0 ( 95% CI 0–2 . 4 ) , and monkeys 0 . 5 ( 95% CI 0–1 . 1 ) ., Likewise most of the rabies positive samples were also from dogs 80 . 0 ( 95%CI 65 . 5–94 . 4 ) , followed by cats 0 . 5 ( 95%CI 0–1 . 4 ) , and other animals 0 . 13 ( 95% CI 0–0 . 4 ) ., These represent that 99 . 2% of the rabies samples were from dogs followed by cats and monkeys ( 0 . 6% and 0 . 3% , respectively ) ., Province-wise ( Table 2 ) data analyses showed that yearly submitted samples were mainly from Vientiane capital 115 . 1 ( 95% CI 90 . 2–140 . 0 ) , followed by Champasak 25 . 4 ( 95% CI 14 . 3–36 . 4 ) , Vientiane province 11 . 7 ( 95% CI 6 . 6–16 . 9 ) , and other provinces 5 . 4 ( 95% CI 3 . 1–7 . 6 ) ., Samples from Vientiane capital represents 73 . 0% of the total submitted samples followed by Champasak ( 16 . 1% ) , Vientiane province ( 11 . 7% ) , and other provinces ( 3 . 4% ) ., From other provinces sample number never exceeded more than six and usually only one per year ., Samples were never received during this period from six of 18 provinces of Laos ., The presence of rabies virus was confirmed in 21 of the 23 FAT positive samples by RT-PCR and partial nucleotide sequence of these amplicons ., Lao9 and Lao19 samples from Vientiane capital were negative by RT-PCR ., The full-length N gene sequence could be done in a total of 15 samples that were used in time-line evolutionary analysis ., These samples were from Vientiane capital and Champasak provinces , the full-length N gene of the only sample from Vientiane province could not be amplified ., The mean rate of nucleotide substitution estimated for the N gene was 2 . 5×104 substitutions/site/year ( 95% HDP values 1 . 6–3 . 4×104 substitutions/site/year ) ., This rate is in agreement with previous studies 11 ., Phylogenetic tree ( Fig 2 ) revealed that rabies strains in Laos belong to the Southeast Asian ( SEA ) cluster that diverged from the Chinese strains approximately 331 . 5 years ( 95% HPD 215 . 1–481 . 9 years ) ago ., From the most recent common ancestor ( TMRCA ) of SEA cluster , approximately 84 . 5 years ( 95% HPD 58 . 8–120 . 7 years ) ago , i . e . , around 1928 ( 95% HPD range 1891 to 1953 ) , rabies virus of Myanmar segregated ., Approximately 62 . 6 years ( 95% HPD 45 . 1–87 . 4 years ) ago , i . e . , around 1949 ( 95% HPD range 1925 to 1967 ) , a cluster only containing Laos rabies viruses segregated from TMRCA of the rest of the SEA cluster which contained viruses from Laos , Cambodia , Vietnam and Thailand ., The Lao rabies viruses segregated approximately 44 . 2 years ( 95% HPD 29 . 0–64 . 1 years ) ago , i . e . , around 1968 ( 95% HPD range 1948 to 1983 ) , into two lineages , one containing viruses circulated during 1999–2002 designated as Laos I , and the other contained viruses we identified in 2011 and 2012 designated as Laos II ., This lineage had 11 viruses only 2 were from Champasak district and rest from Vientiane capital ., Approximately 55 . 3 years ( 95% HPD 40 . 7–76 . 7 years ) ago i ., e ., in 1957 ( 95% HPD 1935–1971 ) Laos III , containing one rabies virus from Vientiane capital and two from Champasak , diverged from rest of SEA cluster which still contained a rabies viruses from Laos ., We designated this lineage as Laos IV which segregated from the other lineages of the SEA cluster , only containing rabies viruses from Cambodia , Vietnam and Thailand , approximately 46 . 4 years ( 95% HPD 35 . 6–62 . 7 years ) ago , i . e . in 1966 ( 95% HPD 1949–1976 ) ., All lineages were supported by significant posterior value ., We completed the whole genome sequence of two strains , strains Lao2 and Lao4 belong to Lao II and Lao III lineage , respectively ., The full genome of both strains was 11924 nt long ., At whole genome level there was 97 . 2% identity between the strains ., The N , P , M , G and L genes , and G–L intergenic region were of same length ( Table 3 ) ., The nucleotide identity of the genes and their deduced amino acid identity were 96 . 2–97 . 6% and 97 . 6–99 . 6% , respectively ( Table 3 ) ., To further identify the different between Lao II and Lao III lineages , strain Lao2 were compared with strain Lao4 to detect the substitutions in the deduced amino acid sequences ( Table 4 ) ., A total of 33 substitutions were identified among them 14 were in various domains of the proteins and 19 were in portion of the protein where no site/domain/region has been identified ., Phylogenetic tree constructed by using whole genome showed that the bat origin rabies viruses from Americas formed a separate cluster from rest of the strains ( Fig 3 ) ., Rest of the strains was divided into China , Southeast Asia , arctic/arctic like , Europe-America and Sri Lanka clusters ., The China cluster consists of two major sub-clusters China I and China II ., All clusters and sub-clusters were supported by bootstrap value of 100 ., Phylogenetic tree revealed that rabies strains from Laos belong to the SEA cluster with Thai strains ., Strain Lao4 from Champasak province was closer to Thai strain than strain Lao2 from Vientiane capital ., There is an increasing trend of rabies virus detection in animal samples in Laos , which may indicate increase of rabies since dog immunization rate is low ., An integrated approach is needed such as public education , access to vaccine and rabies test at affordable cost and time ., Since dogs are main animal reservoir of rabies virus in Laos , therefore it might be easier to control rabies if appropriate program is introduced ., Circulating rabies viruses in Laos are closely related with the rabies viruses from neighboring countries ., Possibly there is no recent invasion of rabies viruses in Laos but in the past there was periodic exchange of rabies viruses with neighboring countries ., Dog movements through Laos for regional dog meat consumption did not cause any spillover recently ., More regional laboratories should be established in different parts of the country for proper surveillance of rabies in Laos . | Background, Methods, Results, Discussion | Although rabies is endemic in Laos , genetic characterization of the viruses in this country is limited ., There are growing concerns that development in the region may have increased transport of dog through Laos for regional dog meat consumption , and that this may cause spillover of the viruses from dogs brought here from other countries ., This study was therefore undertaken to evaluate the current rabies situation and the genetic characteristics of rabies viruses currently circulating in Laos ., We determined the rate of rabies-positive samples by analyzing data from animal samples submitted to the Lao Ministry of Agriculture and Forestry’s National Animal Health Centre rabies laboratory from 2004 through 2011 ., Twenty-three rabies-positive samples were used for viral genetic characterization ., Full genome sequencing was performed on two rabies viruses ., Rabies-positive samples increased substantially from 40 . 5% in 2004 to 60 . 2% in 2009 and continued at this level during the study period ., More than 99% of the samples were from dogs , followed by cats and monkeys ., Phylogenetic analyses showed that three rabies virus lineages belonging to the Southeast Asian cluster are currently circulating in Laos; these are closely related to viruses from Thailand , Cambodia and Vietnam ., Lineages of the circulating Laos rabies viruses diverged from common ancestors as recently as 44 . 2 years and as much as 55 . 3 years ago , indicating periodic virus invasions ., There is an increasing trend of rabies in Laotian animals ., Similar to other rabies-endemic countries , dogs are the main viral reservoir ., Three viral lineages closely related to viruses from neighboring countries are currently circulating in Laos ., Data provide evidence of periodic historic exchanges of the viruses with neighboring countries , but no recent invasion . | Laos is a land-locked rabies-endemic country in Southeast Asia that is surrounded by five rabies-endemic countries ., Thus , there is increasing concern that the epidemiology of rabies in Laos is influenced by infrastructure development and economic activities , including international transport of dogs for meat consumption ., Studies on the epidemiology of rabies are limited in this country ., Therefore , to gain further information about the epidemiology and the genetic characteristics of circulating rabies viruses , this study was conducted using samples submitted to the rabies Lao Ministry of Agriculture and Forestry’s National Animal Health Centre rabies laboratory ., Out of 18 provinces , samples were submitted mainly from the capital Vientiane and Champasak province ., Data from the period 2004 through 2011 showed a gradual increase in rabies-positive samples ., Dogs were the main viral reservoir , and genetic analyses of samples collected from February 2011 through March 2012 showed that three viral lineages are currently circulating in the country ., These rabies viruses are related to those of neighboring countries , indicating shared ancestry but no recent viral invasion . | null | null |
journal.pntd.0005368 | 2,017 | TrypsNetDB: An integrated framework for the functional characterization of trypanosomatid proteins | Trypanosomatid parasites cause life-threatening diseases in humans and major production losses in animals ., They pose global threats , and various issues are associated with available drugs against trypanosomatids ( including tolerability , cost , and resistance ) , necessitating the identification of novel essential parasitic-specific pathways/genes as potential drug targets 1 ., However , as supported by whole genome sequencing data , it is well known that species of the trypanosomatid family , while showing high similarity in proteomes with one another , are highly diverged from other eukaryotes 2–5 ., This makes the annotation transfer of nearly half of their proteome by homology-based approaches from model organisms unreliable 3 ., During the past decade , several genome-wide and focused studies have been conducted to functionally characterize trypanosomatid proteins ., The construction of global and local protein interaction maps has served as one of the main resources for functional annotation by reflecting the molecular context of proteins in a cell 6–15 ., Several experimental techniques exist to identify the interacting partners of proteins that differ in selectivity and sensitivity ., Therefore , one major challenge in the study of protein interactions is the ability to distinguish between the correctly associated proteins from confounding elements that are present in the results of these experiments ., It is also helpful to know the potential interacting proteins that are missing from the results of an experiment based on previously known knowledge of the species of interest or other related trypanosomatid species ., Several databases have been developed to represent the experimentally identified or computationally inferred physical and functional protein interactions 16–21 ., Such databases greatly help researchers to interrogate cellular processes and gain a systems level view of the protein ( s ) of choice ., Although it is of critical importance for studies on trypanosomatids , only a limited number of databases cover information on protein interactions of these parasites , and such interactions are mostly predicted by transferring the available data from other eukaryotes , missing most parts of the published data on trypanosomatid species 16 ., Another major approach for the functional characterization of proteins stems from recent technological advances that have allowed measuring transcriptome , proteome , and transcript half-life changes in response to environmental changes , different life stages , or cell conditions 8 , 22–38 ., Moreover , it is possible to gain insights on the function of a protein by gathering information on:, 1 ) its annotation from resources such as gene ontology , KEGG pathways , and the BioCyc database 37 , 39;, 2 ) protein characteristics , such as protein sequence motifs , isoelectric points , molecular weights , and the number of transmembrane domains;, 3 ) the essentiality of gene knock-down on cell survival 33; and, 4 ) the potential cis-regulatory elements present in the 3′-UTR of the gene and the collective response of genes containing that regulatory element to environmental changes 29 ., Currently , the TriTrypDB database is a gene-centric framework devoted to the kinetoplastid parasites and provides extensive information on the queried protein ranging from genomic sequence and position to involved biological pathways and captured responses in previously reported studies 38 ., However , in many cases , researchers are interested in knowing the collective response of a list of pre-specified proteins along with their interacting partners according to large-scale studies rather than focusing on one protein ., Combining interaction data with enrichment analyses of gene ontology , molecular pathways , gene essentiality , and protein sequence features is the key to perceiving the function of proteins ., Here , we describe TrypsNetDB , a user friendly , integrated database that fills the aforementioned gaps by not only depositing the current interactome knowledge on trypanosomatid proteins but also combining such information with other available resources accompanied with related statistical analyses ., Moreover , the database automatically performs inter-species mapping of the available data and provides information to allow for a better characterization of the queried proteins in the species of interest ., Finally , based on the built-in features , the database can help researchers with their interactome related experiments by distinguishing the likely binding partners of a protein from confounding elements identified in their experiments and suggesting other potentially interacting proteins that are missing from the list of queried proteins ., Built on powerful ASP . Net framework , the database performance is fast and reliable ., TrypsNetDB is freely available at trypsNetDB . org ., The current release of the database is focused on physical protein interaction data that are already published in the trypanosomatid field , supporting 16 trypanosomatid parasites including T . cruzi strain CL Brener , T . cruzi CL Brener Esmeraldo-like , T . cruzi CL Brener Non-Esmeraldo-like , T . brucei gambiense DAL972 , T . brucei Lister strain 427 , T . vivax Y486 , T . evansi strain STIB 805 , T . brucei TREU927 , L . major strain Friedlin , L . mexicana MHOM/GT/2001/U1103 , L . infantum JPCM5 , L . donovani BPK282A1 , L . braziliensis MHOM/BR/75/M2903 , L . braziliensis MHOM/BR/75/M2904 , L . arabica strain LEM1108 , and L . enriettii strain LEM3045 ., Fig 1 represents the schematic architecture of the database ., To systematically extract the protein interaction data , we searched the NCBI PubMed database using the keywords Trypanosoma and Leishmania and extracted all resultant abstracts ., Next , by a manual search , initial positive and negative gold standard sets were constructed by considering 46 and 251 articles , respectively ., A multinomial naïve Bayes classifier was used to prioritize 6581 articles that were more likely to contain protein interaction data based on the abstract content with an estimated probability greater than 0 . 75 ., By a manual inspection of some articles , the initial positive and negative gold standard set was expanded to 60 and 332 articles , respectively ( with extra attention on keeping the diversity of the gold standard sets to reduce the chance of biased predictions ) ., The multinomial naïve Bayes classifier was re-trained and then re-applied to all the extracted abstracts from PubMed using the new gold standard set ., A total of 1996 articles that were likely to include interaction data were identified ( estimated probability of 0 . 9 ) ., By reviewing the articles of the final list , we could extract protein interaction data from 97 different studies ., The interaction data were obtained using a variety of techniques , including affinity purification , immunoprecipitation , yeast two-hybrid ( Y2H ) , fractionation patterns , and other possible experimental techniques ., We have only considered the syntenic orthologs reported by TriTrypDB to transfer the inter-species information ., Users can query the database based on either tritrypDB IDs ( recognizing IDs of recent and older versions of the database ) or gene names ., Support for the remainder of the trypanosomatid species is scheduled to be added in the coming months ., In cases where the gene names match multiple organisms , the user will be asked to select the species of interest from a dropdown box ., As shown in Fig 2 , querying of protein ( s ) will redirect the user to the interaction page , which is composed of the following three main elements: information panel , network , and reference section ., The information panel can be used to explore the details of the three sections of annotations , protein descriptions , and the constructed network ., The annotation tab contains gene set enrichment analyses of the combined set of queried proteins with the suggested proteins by the database ( i . e . , proteins with gray background ) for enrichment in the following five different categories:, 1 ) GO & Pathway: genes are examined for enrichment in Gene Ontology , KEGG pathway , and BioCyc annotations using hypergeometric tests ., Terms with Benjamini-Hochberg corrected p-values less than 0 . 05 are reported back to the user ., Hovering over each term will highlight the proteins that are associated with the term ., Clicking on each represented term will show a description of the term , its category , number of proteins in the network associated with that term , and the corresponding corrected p-value ., 2 ) Sequence & Structure: The sequence and structural features of the proteins are examined , such as the protein motifs , isoelectric points , molecular weights , and predicted number of transmembrane domains ( statistical test for protein motifs is based on the hypergeometric test , and for the other categories based on Wilcoxon-Mann-Whitney rank sum test ) ., This information can provide a complementary view of the function of the proteins ., For example , a group of soluble interacting proteins are expected to have a significantly low number of transmembrane domains ., Likewise , proteins interacting with RNA and DNA are expected to have high isoelectric points ., Similar to the GO and pathway enrichment results , hovering over significantly enriched protein motifs will highlight the associated proteins in the network ., 3 ) Expression patterns: the proteins in the network are examined for their collective transcriptome and proteome responses across 48 distinct samples using Wilcoxon-Mann-Whitney rank sum test ., Each sample is color coded with yellow and blue indicating over-expression/enrichment and under-expression/depletion , respectively ., Statistically significant terms with p-values less than 0 . 05 are highlighted by darker colors , while non-significant conditions are semi-transparent ., The 48 considered cell states were obtained from genome-wide experiments on T . brucei , T . cruzi , and L . infantum 22–24 , 26–28 , 30–32 , 34–36 , 40 , 41 ., By considering syntenic orthologs ( as defined in TriTrypDB ) , the database automatically propagates the information to other trypanosomatid species ., Clicking on each sample will open information on the title of the sample , a description of the results of the statistical tests , the calculated p-value , the title of the study that published the sample and its PMID with a link to the PubMed abstracts ., 4 ) Gene essentiality: The essentiality of proteins in four different cell conditions of T . brucei are examined based by application of hypergeometric tests on the results of a genome-wide phenotyping study 33 ., Ortholog mapping is performed for cases in which the queried organism is not T . brucei ., 5 ) 3′ regulatory elements: Using a novel approach , we recently predicted 88 cis-regulatory elements that are potentially involved in the developmental regulation of T . brucei 29 ., Although only a limited number of functional elements have been identified thus far , by a rigorous analysis of results , we showed that 11 predicted motifs strikingly resemble previously identified regulatory elements in trypanosomatids , suggesting the high accuracy of the predictions ., This section examines whether the 3′-UTRs of the orthologs in the set of proteins in T . brucei are significantly enriched for any of the predicted 88 motifs using hypergeometric tests ., In cases where enrichment is found , the motif logo along with the transcriptome and proteome responses of the motif in different cell conditions are reported ., The proteins tab provides brief information ( such as transcript and protein length , isoelectric point , molecular weight , etc . ) with a link to the TriTrypDB database in a sorted way , starting from the queried proteins and ending with the proteins that were included in the network by the program ( these suggested proteins have been highly connected to the queried proteins based on literature derived interactions ) ., The network tab can be used to explore the contribution of each experimental technique to the construction of the illustrated network ., In two cases of tagged affinity purification and immunoprecipitation in which interactions can show indirect associations , the database distinguishes between interactions that are identified based on RNase treatment of the samples from those that are not ., Hovering over each technique will highlight the interactions that they support ., It is also possible to filter some of the techniques by unchecking the corresponding checkboxes and clicking on the “set filters” button ., The network section , using a dynamically interactive interface , represents the interactions among the proteins with each protein indicated by a circular node ., It is possible to zoom in or out of the network and reposition the proteins ., Queried proteins and other proteins suggested by the database are shown in blue and gray , respectively ., The node size of proteins indicates the number of interactions that they have in the global network with larger nodes representing nodes with a higher number of interactions ., Selecting a protein by clicking on it will highlight the first neighbors of that protein and open the corresponding information in the proteins tab of the information panel ., Finally , the network option on the top-left part of the network section can be used to automatically rearrange the network for perhaps a better presentation or to show/hide the protein labels , which may prove useful for the visualization of relatively large networks ., The reference section provides the references from which the interactions were extracted ., The full source of resources used for the extraction of the interaction data can be accessed by going to the “References” section from top menu or going directly to the trypsNetDB . org/references . aspx webpage ., The “Genome-wide data” section on the menu enables users to visualize the genome-wide data available for the queried proteins and their interacting partners suggested by the database ., The supported genome-wide data in the current release of the database are categorized in three main groups of fractionation patterns , gene expression patterns , and phenotypic effects , with each containing sub-categories ., Users can select one of the main categories ( indicated with a blue background ) to represent all related sub-categories at once or directly select the sub-categories ., This part of the database is particularly useful for the validation of results obtained from interactome-related experiments ( such as affinity purifications ) by helping users to distinguish between direct binding partners of a protein from potentially spurious elements ., For example , the fractionation heatmaps can be exploited to assess whether the potentially interacting proteins show similar fractionation patterns ( Fig 3 ) ., Currently , the database provides fractionation patterns for whole-cell , mitochondrial-enriched , and cytosolic-enriched cell extracts that can be informative for the localization of previously unannotated proteins ( i . e . , mitochondrial proteins are expected to be identified in the mitochondrial-enriched fractions while depleted in the cytosolic-enriched sample ) ., Fractionation patterns are also informative regarding the nature of the interactions ., As described elsewhere 8 , glycerol gradient-based fractionation patterns can capture more transient interactions , while ion exchange-based fractionation favors more stable interactions due to the presence of a salt gradient ., Finally , physically interacting proteins are expected to be involved in similar biological processes and , hence , show similar expression patterns and degrees of essentiality in each cell state , which can be easily assessed using the corresponding heatmaps ., By going to the save option on the top of the network section , users can save the whole represented network or only the sub-networks that are supported by a specific experimental technique ., It is also possible to save the enrichment analysis results and the annotations of genes , such as the description , transcript or protein characteristics ( length , weight , isoelectric point , and identified SNPs ) , and gene ontology ., Users can also use the save query list option for later regeneration of the same results ., The web application is developed based on the . Net framework 4 . 5 technology ., To improve the performance , the statistical analysis modules ( including hypergeometric test , Wilcoxon-Mann-Whitney test , and Benjamini-Hochberg p-value adjustment procedure ) were implemented in C# and added as a library to the web application and the performance of the modules has been validated by comparing the results with those of MATLAB 2015b on multiple test sets to ensure accuracy ., The network visualization is based on the cytoscape web library , which requires flash player for the representation of the network ., All analyses are performed in real-time and a session for each user is ended after 1hr of inactivity ., For high-performance , the database is implemented in Microsoft SQL Server 2012 ., Protein interaction maps remains one of the major resources for the functional annotation of proteins ., Embedding other lines of information with these maps can help researchers gain insights regarding the molecular contexts of the proteins ., Here , we introduce TrypsNetDB , a web tool to consolidate the current knowledge on the interactome of the trypanosomatid parasites and dynamically integrate them with a wealth of available orthogonal information ., We are continuously working on expanding the core , literature-derived , protein interaction depository of the database ., Future plans also include providing supports for the remaining trypanosomatid parasites and the inclusion of other genome-wide data ., TrypsNetDB is an open source effort , and hence , the code and databases are available through the portal ., Moreover , the interaction and fractionation data can be directly downloaded from the web interface using the provided links . | Introduction, Program description and methods, Conclusions and future directions | Trypanosomatid parasites cause serious infections in humans and production losses in livestock ., Due to the high divergence from other eukaryotes , such as humans and model organisms , the functional roles of many trypanosomatid proteins cannot be predicted by homology-based methods , rendering a significant portion of their proteins as uncharacterized ., Recent technological advances have led to the availability of multiple systematic and genome-wide datasets on trypanosomatid parasites that are informative regarding the biological role ( s ) of their proteins ., Here , we report TrypsNetDB ( http://trypsNetDB . org ) , a web-based resource for the functional annotation of 16 different species/strains of trypanosomatid parasites ., The database not only visualizes the network context of the queried protein ( s ) in an intuitive way but also examines the response of the represented network in more than 50 different biological contexts and its enrichment for various biological terms and pathways , protein sequence signatures , and potential RNA regulatory elements ., The interactome core of the database , as of Jan 23 , 2017 , contains 101 , 187 interactions among 13 , 395 trypanosomatid proteins inferred from 97 genome-wide and focused studies on the interactome of these organisms . | Methods to predict protein function based on sequences enable the rapid annotation of newly sequenced genomes ., However , as most of these methods rely on homology-based approaches , non-conserved proteins in trypanosomatids remain elusive for annotation , rendering approximately half of the sequenced proteins uncharacterized ., In this study , we developed a user friendly integrated database , TrypsNetDB , which fills multiple gaps in the field by depositing the current interactome knowledge on trypanosomatid proteins and combining this information with other available resources accompanied by related statistical analyses ., The database allows automatic inter-species mapping of available data to better characterize the queried proteins in the species of interest ., The database is built on fast and reliable ASP . Net framework and provides, ( i ) a significant increase in the genome-wide functional annotation of trypanosomatid proteins ,, ( ii ) potential novel targets for therapeutics against trypanosomatids , and, ( iii ) a robust methodology that can be adapted for the functional annotation of other non-model organisms . | protein interactions, protein interaction networks, parasitic protozoans, organisms, genomic databases, trypanosoma brucei gambiense, protozoans, network analysis, genome analysis, sequence motif analysis, research and analysis methods, separation processes, sequence analysis, computer and information sciences, bioinformatics, proteins, biological databases, proteomics, biochemistry, trypanosoma, sequence databases, data visualization, database and informatics methods, genetics, biology and life sciences, genomics, computational biology, fractionation | null |
journal.pcbi.1000396 | 2,009 | The Dynamics of T-Cell Receptor Repertoire Diversity Following Thymus
Transplantation for DiGeorge Anomaly | The steady state T-cell population size arises in the balance among several, phenomena , including the rapid and extensive expansion of rare clones through, activation-induced peripheral division and their subsequent contraction ( which, constitute , in part , the T-cell immune response ) and the relatively slow, turnover of a diverse pool of naive cells through continuous thymic emigration, and cell death ., The specific memory T cells that arise as the result of clonal, expansion appear to be regulated largely independently of the naive cells 11 ., T, cells arising through lymphopenia-induced proliferation acquire markers that, ordinarily indicate a memory phenotype and may be regulated as memory cells ,, though this hypothesis has not been definitively tested ., Both the size and diversity of peripheral T cell populations are controlled, through competition for limiting resources ., ( In the interest of simplicity , we, will use the term “resources” familiar from ecological, population studies to refer to all of the factors that mediate growth, regulation ., We do not intend by this to exclude factors that are more accurately, referred to as “signals” ) 5 , 13 ,, and may also be controlled directly by the activities of regulatory T cells, 14 ., It has been shown , for example , that normal, T-cell population growth is dependent on stimulation by self-peptide major, histocompatibility complex ( spMHC ) complexes through the TCR 15–18 , requiring a TCR, that is specific for the spMHC complex ., But T-cell population growth also, depends on cytokines such as IL7 and IL15 that act independently of TCR, specificity 5 , 19–21 ., Furthermore , growth and survival in all cells require adequate space and, nutrients , the utilization of which is independent of TCR specificity 22 , 23 ., Our current understanding of lymphopenia-induced proliferation is due to studies, in mice demonstrating that T cells divide rapidly after transfer into T, cell-deficient ( usually due to RAG or CD3 deficiency ) or irradiated mice , but not after transfer to, normal mice 15 , 16 ., Moreover ,, overall T cell numbers in T cell-deficient animals after transfer and clonal, expansion are similar to T cell numbers in normal animals , suggesting control, mechanisms acting on total T cell numbers , rather than in a clone-specific, manner ., The importance of TCR specific signals has been studied at length , showing that, competition within T cell clones is important in maintaining TCR repertoire, diversity ., It has been shown , for example , that in a T-cell-deficient host , a T, cell must interact with antigen-presenting cells bearing the MHC allele, responsible for that cells thymic selection in order to proliferate, 24 ., In a T-cell sufficient host , such TCR-spMHC, interaction is necessary for T cell survival 24 , 25 ., Furthermore , naive polyclonal T cells divide when transferred to TCR-transgenic, hosts , as do monoclonal T cells transferred to TCR-transgenic hosts of differing, clonotype ., T cells do not divide , however , in hosts of identical clonotype 26 ., Mice, lacking MHC class II expression do not repopulate the periphery with CD4 T cells, at all , suggesting that peripheral MHC class II expression is needed for the, survival of CD4 T cells 25 ., In the present context it is important to, note that MHC class II matching in thymic grafts for complete DiGeorge subjects, is not necessary for the development of CD4 T cells 27 ., TCR-nonspecific signals include cytokines such as the cytokine interleukin-7, ( IL7 ) , which is necessary for the survival of nave T cells 28–32 ., Lymphopenia-induced proliferation of memory cells requires IL7 or IL15 13 , 33 ., T cells that, have lost the ability to respond to IL7 after leaving the thymus are no longer, able to proliferate , produce cytokines , or acquire memory cell phenotype 30 ., In, mice in which IL7 signaling has been completely abrogated , the few mature T, cells found in the peripheral blood behave abnormally 30 , 34 , 35 ., Experiments in, IL7-receptor ( ) -deficient mice ( IL7R−/− ) have, shown a reduction in T-cell capacity to proliferate upon stimulation , leading to, a six- to seven-fold reduction in the frequency of clonogenic T cells compared, with T cells from IL7R-sufficient mice ( IL7R+/+ ) ,, as well as a 50% reduction in the average clone size of single, IL7R−/− T cells compared with the, IL7R+/+ T cells 34 ., In another, study , mice lacking the interleukin 2 ( IL2 ) receptor chain ( ) and/or the Jak family tyrosine kinase ( Jak-3 ) had severe, combined immune defects with lack of T lymphocyte maturation and function ., This, phenomenon is presumably attributable to the fact that is part of the receptor for IL7 and IL15 36; its loss leads to, the abrogation of both of these cytokine signaling pathways 30 , 35 , and others as well ., Moreover , in humans , in the absence of of or of the Jak-3 family , the periphery lacks T cells completely, 37 , 38 ., To assess the contributions of thymic emigration rate on the steady-state T-cell, population size , Berzins et al 39 engrafted variable numbers of thymuses into, mice and observed that the size of the naive T-cell population increased in, proportion to the number of thymic grafts , while the size of the memory, population remained unchanged ., It may be of importance to note that the thymus, grafts themselves produced IL7 ., In humans , transplantation of thymic tissue at, varying doses into complete DiGeorge anomaly subjects showed no significant, effect on the nave CD4 or CD8 T cell numbers 27 ., Complete DiGeorge Anomaly ( cDGA ) is a congenital condition , the hallmark of which, is profound immunodeficiency arising from abnormal development of the third and, fourth pharyngeal pouches resulting in thymic aplasia ( complete absence of the, thymus ) ., This developmental irregularity may also cause various other anatomical, abnormalities including heart defects , hypoparathyroidism , and craniofacial, malformations 40–42 ., In the absence of a, thymus , cDGA subjects lack thymic-derived T cells and are consequently, profoundly immunocompromised ., Postnatal thymus transplantation can lead to a, restoration of T-cell function and the development of a peripheral T-cell, population with an apparently normal T-cell receptor repertoire 43–45 ., Thymus transplantation is emerging as a valuable treatment for athymia generally ., There is therefore substantial medical interest in elucidating the immunological, recovery of thymus transplant patients quite apart from the basic immunology, these patients may reveal ., To illustrate the connection between these alternative homeostatic mechanisms and, the observations one might make on the DiGeorge subjects , consider the following, two scenarios as a thought experiment ., First , suppose that TCR-specific, resources such as spMHC are not limiting , either because they are produced in, great excess or because they are rendered unnecessary ., Under these conditions ,, homeostasis will be due exclusively to competition for TCR-nonspecific, resources ., The first T cells leaving the thymus will expand rapidly , consuming, the TCR-nonspecific resources required for growth of their own growth and for, the growth of all other clones ., Subsequent T cells leaving the thymus will, encounter a more impoverished environment and will grow more slowly , leading to, early dominance of one or a small number of early clones and therefore a limited, TCR repertoire ( Figure 1 ) ., Conversely , if TCR-nonspecific resources such as cytokines were not limiting ,, and that homeostasis therefore depended solely on competition for TCR-specific, signaling , the only cellular competition would be among T cells of the same, clonotype ., In this case , each clone will grow to roughly the same self-limiting, size regardless of when its founder emigrates from the thymus ., In this case , TCR, repertoire diversity will grow at the greatest possible rate , all other things, being equal ( Figure 1 ) ., Reality will lie somewhere between these two limiting cases: TCR repertoire, diversity will be shaped by both TCR-specific and TCR-nonspecific resources ., Our, goal is to explore , quantitatively , how these two sets of signals interact, dynamically to shape the mature functional T cell population ., The quantitative contributions of different signals in the regulation of T-cell, number and the diversity of the TCR repertoire are very difficult to determine, experimentally in the context of an intact system ., To examine the interplay, among the mechanisms responsible for the heterogeneity of the T-cell repertoire ,, we have developed a mathematical model for the effective interactions among, T-cell clones ., We model the temporal evolution of T-cell clones and their, dynamics under the combined effect of TCR-specific and TCR-nonspecific signals ., In particular , we consider competition within clones for spMHC complexes, presented by antigen-presenting cells as well as competition among cells in all, clones for cytokines , space and other TCR-non-specific resources ., Such mathematical models have been used extensively to study the dynamical, interactions between different branches of the immune system 46 and, their effect on T cell 47–51 and B cell, population development 52 ., In this paper , we develop a model to examine, competitive interactions among homo- and heterospecific T cells under limiting, spMHC and TCR-nonspecific resources ., This model then provides the basis for the, analysis of clinical data from complete DiGeorge subjects following thymic, transplantation to estimate the extent to which regulatory mechanisms working, through these distinct pathways contribute to shaping the T cell population ., The, distinguishing feature of our model is its explicit representation of individual, T cell clones , and the direct focus on the interplay between T-cell population, size and TCR repertoire diversity this strategy allows ., All model parameters are, fit to data gathered within the present study; there are no parameters estimated, from external sources ., The Kullback-Leibler divergence ( ) is a measure of the difference between two probability mass, functions that arises naturally in the context of likelihood ratio tests for, multinomial distributions ., The diversity of patient TCR repertoires can be, quantified via -based comparison of patient spectratypes to mean spectratypes, averaged over healthy controls 55 , and -based comparison of TCRBV frequencies to those in healthy, controls ., Furthermore , can be decomposed hierarchically into components attributable ,, respectively , to variation in TCRBV family usage , CDR3 length variation within, TCRBV family , and protein sequence variation within CDR3 length ., For this work ,, we measure TCRBV family usage by flow cytometry and distributions of CDR3, lengths by spectratyping ., The fact that we are not analyzing sequence data, implies that we are not making the highest-resolution diversity measurements, possible ., We will show , however , that this potential limitation does not impede, our ability to perform the measurements we set out to make ., The data obtained in, each of these assays represents the relative frequency of TCR counts in a, subclass conditional on the parent class: is the fraction of cells using a gene segment from TCRBV, family in a given DiGeorge subject and is the mean usage in our collection of healthy controls ., , and are the relative numbers of T cells of CDR3 lengths within those cells using TCRBV family in a given patient , and among controls , respectively ., The at the level of TCRBV is ( 6 ) where is the number of TCRBV families considered , which in our case, is 23 ., The CDR3-length divergence within the family is ( 7 ) where is the number of CDR3 lengths , in our case 18 ., The total, divergence is ( 8 ) The inverse of the can be regarded as the “completeness” of a, TCR repertoire with respect to some standard 55 ., Under very, reasonable assumptions , the completeness may be regarded as a measure of the, diversity , so we will refer to the diversity of the TCR repertoire , and quantify, it by the sample completeness , ., Note that diversity increases as the the decreases toward its minimum value , zero ., The sample , like the sample entropy , is a biased estimator of the, population : the expected value of the sample differs from the population by an additive term inversely proportional to the sample size ., We account for this fact by including an additive bias as a parameter in the estimation procedure ., We can then compare the estimated, bias against the sample sizes ( Table 1 ) for each datum ., The model is based on absolute T-cell numbers , but the data are blood T cell, concentrations ., We must use a conversion factor that converts total T cell numbers to T-cell blood, concentration ., This factor is not easy to estimate with any precision , but our, results are quite insensitive to large variation in ., Assuming that a 10 kg subject has 600 ml of blood , and that there are 45 times, more lymphocytes in the tissues than in the blood , based on adult data 56 , we, have ., This is the conversion factor we use in the rest of the, paper ., Furthermore , some subjects have a relatively small number of anomalous T cells at, the time of transplantation ., We denote the concentration of such cells and treat them as a separate non-functional family ., The model, quantity ultimately used for comparison to patient T-cell concentration data is ., We estimate the remaining parameters , and , as well as the measurement error variances and , and the measurement bias β , by fitting the, model specified by Eq ., ( 1 ) to the patient T cell concentration and TCR diversity, data simultaneously using Markov Chain Monte Carlo ( MCMC; see , eg , 57 ) ., We estimate parameters by using the likelihood function given by ( 9 ) where and are the estimated T cell concentrations and diversities ,, respectively , under the model specified by Eqs ., ( 1 , 4 ) ; and are the corresponding patient data observed at sampling times and , respectively ., and are the total number of T cell concentration and diversity, data points , respectively; and are the error variances ., We compute the posterior joint, density of all model parameters using a block-sampled Metropolis-Hastings MCMC ., The parameters are all manifestly positive , and ., Therefore , we transformed , and using natural logs ., was treated using a logistic transformation ., represents the maximum growth rate of T cell populations ., The, maximum possible rate is set by the minimum cell-cycle time for lymphocytes of, about 6 hours , so we likewise used a logistic transformation on ., The prior distributions on the parameters were uniform in the, transformed variables ., The maximum-likelihood parameters estimates are shown in Table 2 , along with the 95%, credible intervals ., The estimate for was much larger than the largest observation time , so we, subsequently fixed , thereby eliminating one modeling degree of freedom and refit, the model ., We found that this simplification entailed no significant decrease in, likelihood for any data set ., This result implies that our experimental design is, not capable of resolving the full time-course of the grafts becoming, functional , not necessarily that the emigration rate continues to increase, indefinitely ., The estimate of greatest interest for us is , which we find lies between about 10−4, and 10−3 in these three subjects ., In subject one , who has, the most complete data available and the tightest posterior marginal, distribution on , produces an estimate midway between the other two ., The parameter governing total thymic emigration , , does not have a simple direct interpretation , but one can, estimate the thymic emigration rate at any time post-transplantation ., These, rates , estimated for the three subjects at one year post-transplantation , are, 93 , 44 , and 198 cells per day , respectively ., These rates are several orders of, magnitude smaller than those estimated for healthy humans infants ., We can examine the estimated values of the the bias and use the known expression for this bias to compare to, the actual sample sizes ., Where the underlying distribution is multinomial , the, theoretical bias is given by where is the number of classes in the multinomial , and is the sample size ., Using this formula to estimate the sample, size leads to the conclusion that the effective sizes of the samples used to, estimate diversity are between 103 and 105 cells ,, consistent with the measured sample sizes ( Table 1 ) ., Figure 3 provides a visual, assessment of the quality of the model fits to these data ., In order to gain a deeper understanding of the role of on the dynamics of both T cell population size and diversity ,, we performed two additional analyses ., We use the data from subject one to, illustrate these points ., Analyses on the other two subjects yielded comparable, results and lead to the same conclusions ., A reasonable concern is that thymic emigration rate and may be confounded–that it may be possible to, compensate for smaller by making larger ., We addressed that concern by performing a numerical, experiment to determine the impact on the model fit of fixing outside the inferred credible bounds ., We fix to be 10-fold smaller or larger than its maximum-likelihood, estimate for subject 1 ., The fits are significantly worse ( compare Figures 3 and 4 ) , with log-likelihood ratios, of about 8 in both cases ( Table, 3 ) ., A focused analysis of the time-dependent sensitivity of the model trajectories to, parameter variation may provide further insight into the mechanisms of, regulation ., We examined the sensitivity of our model to changes in, 58 , 59 and as follows ., Define the sensitivity functions , and ., The sensitivity equations are obtained by differentiating, both sides of equation ( 1 ) with respect to ., The are computed as the solutions of ( 10 ) with for all i ., The sensitivity function for, D is given by ., The relative sensitivities are calculated by multiplying the, absolute sensitivity by and dividing by the value of the relevant response variable ., The sensitivity to is due largely to the changed timing of discrete events , and, so does not lend itself easily to differential sensitivity analysis in the form, outlined in Eq ( 10 ) ., Sensitivity to was therefore estimated directly by generating sample paths, under different values of ., We know that mean sensitivity to is transient because the steady-state population size and, diversity do not depend on in the deterministic limit ( Eqs . 2 , 3 ) ., We find , however , that, the diversity remains sensitive to changes in throughout the two-year period of the study ( Figure 5 ) ., It should be noted ,, however , that this assessment is based on the true diversity rather than the, sample diversity , whose resolution is limited by sample size ., At the sample, sizes available , the sensitivity of the measurable diversity to vanishes between 6 and 12 months ., The population size is very, sensitive to in the first 3–4 months , becoming less so rapidly, after that time ., As increases , the population size decreases during the sensitive, period ., Increases in the thymic emigration rate have a positive effect on both the population size and the, diversity , as expected ., In this case , the sensitivity of the diversity is, greatest earlier than in the case of ρ ., It should be, noted that diversity at the level of amino acid sequence differences is not, resolved by our assays , regardless of sample size ., If it were , we would expect, the diversity to remain sensitive to both of these parameters for a longer, period of time ., The model does not account for the micro-environmental perturbations invariably, encountered by people , including our study subjects ., Infection produces a, transient change in the steady state of the T cell repertoire , typified by a, decrease in the diversity ., Once the infection has been resolved , the T-cell, population returns to its steady state , and the diversity relaxes back to its, original value ., The rate at which the return to steady state values occurs, depends on ρ ., To examine this dependence , we carried, out the following numerical experiment ., We start with a system at steady state , and suddenly increase the size of a, single clone by a factor of 104 , allowing it to consume the non-TCR, specific resources its new size requires ., The diversity decreases as a result ., Over a few days , the other clones adjust to the new steady state ., After ten days, we remove the artificial support for the enlarged clone and allow the system to, return to steady state ( Figure, 6 ) ., We ran this experiment using models with four values of, ρ differing over four orders of magnitude ., For high, ρ , corresponding to a high competition for specific, signals , the diversity decreases less than in the absence of intra-clonal, competition ., Moreover , after the perturbation is resolved , the diversity, increases back to its steady-state , pre-perturbation value at a rate that, depends very sensitively on ., The return to steady-state diversity is more rapid the, greater the competition for TCR-specific signals ., The estimated values for obtained from the DiGeorge patients implies a diversity-return, time on the order of a few days , rather than weeks or more ., The functional integration of the manifold processes involved in the development and, maintenance of the mature T cell repertoire has not yet been fully elucidated ., These, processes include thymic export , competition for TCR ligands , and competition for, non-specific stimulatory factors ., Here , we study T cell homeostasis in complete, DiGeorge Anomaly patients during the establishment of T cells following thymic, transplantation ., To quantify the balance between TCR-specific and TCR-nonspecific factors that act to, limit T-cell population growth , we developed a mathematical model that accounts for, intra- and inter-clonal competition ., The key parameter in this model for the present, purpose is ρ ., This parameter can be given a functional, interpretation by considering the case of an individual capable of rearranging just, a single TCR ., The size of this hypothetical sole clone , at equilibrium , must be, larger than the size of the same clone in the presence of a T cell population with a, complete repertoire ., The ratio of these two sizes is, 1∶ρ ., We have estimated to be about 10−3; the carrying capacity for a, single isolated clone is about 1000 times larger that that for the same clone under, normal heterogeneous conditions ., This result suggests that TCR-specific resource, limitation is relatively unimportant , but this is true only for the total T-cell, population size near steady state; away from steady state , TCR-specific regulation, may is crucial to the establishment and maintenance of repertoire diversity ., At steady-state , TCR diversity is maintained through competition for self peptide MHC, 60 , and population numbers are maintained through, competition for cytokines and other resources not specific to the TCR 5 , 19 ., Our models fit the data adequately only when the thymic emigration is an accelerating, function of time through the initial post-transplantation phase , as would be, expected from the biology of thymus transplantation 61 , 62 ., The slices of, cultured thymus tissue used for transplantation are cultured from 2–3, weeks prior to transplantation ., There is dramatic depletion of thymocytes from the, tissue , thymic epithelium becomes condensed , and the cortico-medullary distinction, is lost ., In the first months after transplantation , the epithelium differentiates so, that cortex and medulla again are distinct and the epithelium develops its, characteristic lacy appearance ., During this process , the numbers of thymocytes in, the transplanted tissue increases dramatically , from small numbers of scattered, thymocytes to densely packed thymocytes throughout the allograft ., Thus , the biology, predicts an increase in emigration with time after transplantation as normal thymic, architecture is reestablished ., The current study was not able to treat naive and memory T cells separately , but, doing so is clearly of great interest given the indications that these pools are, regulated independently of each other 11 ., Furthermore , we were, not able to measure thymic emigration directly , though these measurements would be, of significant interest given the surprisingly small rates inferred here , and the, impact that the reconciliation of this implied inconsistency would have on our, understanding of the treatment of athymia and thymic dysfunction ., Finally , our model, contains terms for the density-dependent competition among T cells ( Eq . 1 ) whose, forms are commonly accepted in population modeling but do not have direct empirical, justification in the present context ., Although we believe that our conclusions are, robust against reasonable variation in these terms , it would be of great utility to, explore this issue explicitly ., Our aim has been to study T cell homeostasis in a more natural human system than has, otherwise been available thus far ., It remains to be shown just how natural the, post-transplantation environment is in this regard , and therefore how broadly, applicable our findings are ., We are encouraged , however , by the fact that complete, DiGeorge patients receiving thymus transplants recover adaptive immune function ., It, is important to note that total T-cell counts in these recovered patients remains, below normal , typically at about the 10th percentile for children of the same age, 45 ., Although specific reasons for this condition have not yet been determined , no, molecular defects in T-cell function have been reported ., We regard it as possible, that the sole immunological lesion in DiGeorge anomaly is athymia , and that , once, ameliorated by transplantation , the emergence of the T cell population proceeds as, it would in a non-affected T-cell lymphopenic subject ., We expect , therefore , that, our findings will be generally applicable ., Our study enabled us to discriminate between two qualitatively different forms of, population regulation: TCR-specific regulation and TCR-nonspecific regulation , and, to elucidate their roles in jointly maintaining the dynamic homeostasis of the T, cell population ., The parameters estimates obtained by analyzing data from three, DiGeorge Anomaly patients suggest that T-cell population size is maintained by, TCR-nonspecific mechanisms , and TCR repertoire diversity is maintained by, TCR-specific mechanisms . | Introduction, Model, Results, Discussion | T cell populations are regulated both by signals specific to the T-cell receptor, ( TCR ) and by signals and resources , such as cytokines and space , that act, independently of TCR specificity ., Although it has been demonstrated that, disruption of either of these pathways has a profound effect on T-cell, development , we do not yet have an understanding of the dynamical interactions, of these pathways in their joint shaping of the T cell repertoire ., Complete, DiGeorge Anomaly is a developmental abnormality that results in the failure of, the thymus to develop , absence of T cells , and profound immune deficiency ., After, receiving thymic tissue grafts , patients suffering from DiGeorge anomaly develop, T cells derived from their own precursors but matured in the donor tissue ., We, followed three DiGeorge patients after thymus transplantation to utilize the, remarkable opportunity these subjects provide to elucidate human T-cell, developmental regulation ., Our goal is the determination of the respective roles, of TCR-specific vs . TCR-nonspecific regulatory signals in the growth of these, emerging T-cell populations ., During the course of the study , we measured, peripheral blood T-cell concentrations , TCRβ V, gene-segment usage and CDR3-length spectratypes over two years or more for each, of the subjects ., We find , through statistical analysis based on a novel, stochastic population-dynamic T-cell model , that the carrying capacity, corresponding to TCR-specific resources is approximately 1000-fold larger than, that of TCR-nonspecific resources , implying that the size of the peripheral, T-cell pool at steady state is determined almost entirely by TCR-nonspecific, mechanisms ., Nevertheless , the diversity of the TCR repertoire depends crucially, on TCR-specific regulation ., The estimated strength of this TCR-specific, regulation is sufficient to ensure rapid establishment of TCR repertoire, diversity in the early phase of T cell population growth , and to maintain TCR, repertoire diversity in the face of substantial clonal expansion-induced, perturbation from the steady state . | Protective adaptive immunity depends crucially on the enormous diversity of the, T-cell receptor repertoire , the antigen receptors expressed collectively on, T-cell populations ., T cells develop from T-cell precursors that originate in the, bone marrow and migrate to the thymus , where their T cell receptors are, constructed stochastically , and tested for autoreactivity against a host of self, antigens ., Complete DiGeorge anomaly is a rare congenital disease in which the, thymus fails to develop , blocking all T cell development and causing profound, immunodeficiency ., Thymus transplantation , performed in the first two post-natal, years , allows the patients own T cell precursors to develop in the, engrafted thymus tissue into normal , functioning T cells ., In addition to saving, patients lives , this procedure provides an extraordinary opportunity, to study the de novo development of human T cell populations ., We have developed a mathematical model to aid in the statistical analysis of the, precious data from these patients ., In addition to helping elucidate the means by, which the size and diversity of T cell populations are jointly regulated , the, insights gained from this study hold promise for the development of therapies to, promote immune recovery after transplantation . | mathematics/statistics, immunology/leukocyte development | null |
journal.pcbi.1007293 | 2,019 | TeXP: Deconvolving the effects of pervasive and autonomous transcription of transposable elements | Long interspersed nuclear element 1 ( LINE-1 ) has attracted much attention in the last decade due to its capacity to promote genetic plasticity of the human genome ., LINE-1 is a DNA sequence capable of duplicating itself and other DNA sequences by mobilizing messenger RNAs ( mRNAs ) to new genomic locations via retrotransposition 1–3 ., There are multiple molecular mechanisms to deactivate LINE-1 instances , most prominently , the truncation of 5’UTR due to partial retrotransposition has resulted in mostly inactive and truncated copies of LINE-1 across the human genome3–6 ., Truncated copies of LINE-1 lack their internal promoter sequence and therefore , are expected to be dead-on-arrival ., Although full-length LINE-1 activity has been described in both healthy and pathogenic tissues 3 , 7 , 8 , quantifying its activity is remarkably difficult due to its repetitive nature ., Until recently , LINE-1 retrotransposition was believed to occur in germ cells 9–11 and tumors 12–14 , but not in somatic tissues ., However , growing evidence suggests that LINE-1 is active in the neuroprogenitor cells and in other healthy somatic tissue at low levels 15–18 ., As opposed to healthy tissues , tumor and tumor derived cell lines show higher levels of LINE-1 activity 13 ., LINE-1 instances are likely to be activated due to broad demethylation of LINE-1 promoter 19 ., The literature describes many other factors contributing to the constraints of LINE-1 activity pre- and post-transcriptionally 20; however , little is known about its activation and impact in tumors 21 ., A major challenge to asses LINE-1 activity is the requirement of either specialized assays 22 , 23 or multiple and complementary datasets 24 , hindering estimation of autonomous LINE-1 transcription in a large number of samples ., Moreover , affordable methods to quantify LINE-1 activity , such as those based on RNA 17 , 25 , 26 , are largely confounded by the high copy number nature of LINE-1 and pervasive transcription 23 , which refers to the idea that the majority of the genome is transcribed , beyond just the boundaries of known genes 27 ., How much pervasive transcription influences the human transcriptome is still unclear 27–29 ., Some researchers suggest that pervasive transcription is mostly derived from technical and biological noise and , therefore , might not be relevant in RNA sequencing experiments 30 ., Others suggest that pervasive transcription has a stochastic nature , and if sequenced at enough depth the majority of the genome may be transcribed ., With either theory , pervasive transcription should not affect quantification of the transcription of protein coding genes , which are present either in single copy or low copy numbers in the genome ., However , the quantification of the transcriptional activity of transposable elements , including LINE-1 , would be greatly affected by pervasive transcription due to their multi-copy nature ., The autonomous transcription of LINE-1 , on the other hand , derives from LINE-1 transcripts being fully transcribed from its internal promoter ., Thus , by definition , since LINE-1 promoters are at the 5’ extremity of LINE-1 elements , autonomous transcription is more likely to derive from full length LINE-1 instances ., These transcripts could derive from both intronic or intergenic full-length LINE-1 instances ., This paper presents a new method to remove the effect of pervasive transcription on RNA sequencing datasets and reliably quantify LINE-1 subfamily transcriptional activity ., We first show that the vast majority of reads overlapping LINE-1 elements are derived from pervasive transcription and propose a method to address this issue ., We validated the LINE-1 transcription landscape in well-established human cell lines and their cell compartments ., Finally , we surveyed LINE-1 activity in a variety of healthy somatic tissues ., Although somatic retrotransposition has been mainly studied in the human brain , we found surprisingly little transcriptional activity in most brain regions from adults ., Instead , we found LINE-1 transcriptional activity in other somatic tissues consistent with an overall trend of LINE-1 activity in cell with higher turnover ., We benchmarked TeXP by estimating the autonomous transcription of LINE-1 subfamilies in RNA sequencing experiments of well-established human cell lines 31 ., Fig 2A shows the proportion of reads mapped to LINE-1 subfamilies using a naïve method ( left panel ) and proportions of reads from each signature using TeXP ( right panel ) ., TeXP estimations were also compared to other transposable element quantification pipelines ( S2 Fig ) ., In the naïve method ( Fig 2A; left panel ) , cytoplasmic and whole-cell polyadenylated ( polyA ) + samples had an enrichment of reads mapping to L1Hs and L1PA2 when compared to whole-cell transcripts without a polyadenylated tail ( whole-cell polyA- ) and nuclear RNA samples ., The enrichment of L1Hs reads was consistent with increased transcription of full-length L1Hs ( S3 Fig ) ., The estimates after applying TeXP ( Fig 2A; right panel ) revealed two major signals in MCF-7 RNA sequencing experiments: pervasive transcription and L1Hs autonomous transcription ., The difference between the naïve method and TeXP suggests that reads mapped to ancient LINE-1 subfamilies , such as L1PA3 and L1PA4 , are mostly derived from pervasive transcription ., TeXP also detected residual L1PA2 transcription in a small number of samples ( Fig 2A and S4 Fig ) ., This result is consistent with L1Hs and L1PA2 being the only two LINE-1 subfamilies capable of autonomous transcription and autonomous mobilization in the human germline and tumors 11 , 33 ., MCF-7 , a cell line derived from breast cancer , was previously described as having remarkably high levels of L1Hs autonomous transcription 17 , 24 ., The transcriptome of MCF-7 and many other cell lines were carefully and consistently sequenced through the Encyclopedia of DNA elements ( ENCODE ) project ., Leveraging these ENCODE cell line datasets , we assessed L1Hs autonomous transcription in distinct cell compartments ( S5 Fig and S6 Fig ) 31 ., First , we found that MCF-7 whole-cell polyA+ samples had extremely high levels of L1Hs transcription ( 180 . 7 RPKM ) , in agreement with the literature ., Selecting whole-cell polyA- samples reduced the signal of L1Hs autonomous transcription by 73% ( Fig 2A ) , suggesting that most of the signal was derived from mature polyA+ LINE-1 transcripts ., Furthermore , we tested whether L1Hs transcripts are derived from cytoplasmic ( mature ) or nuclear ( pre-mRNA ) portions of the cell ., We found that nuclear transcripts were highly enriched for pervasive transcription ( autonomous/pervasive ratio 0 . 02 ) , whereas cytoplasmic transcripts had an autonomous/pervasive ratio similar to transcripts derived from whole-cell polyA+ samples ( 0 . 45 and 0 . 51 , respectively–Fig 2A ) ., Together , these results suggest that most of the LINE-1 autonomous transcription signal is derived from mature transcripts in the cytoplasm and only a small fraction of signal is derived from fragmented LINE-1 transcripts in the nucleus ., Analyzing other lymphoblastic and cancer-derived cell lines such as GM12878 , SK-MEL-5 and K-562 yielded no evidence of L1Hs autonomous transcription in most cell compartments or RNA fractions , despite low levels of L1Hs autonomous transcription in whole-cell polyA+ samples ( 0 , 8 . 8 and 8 . 4 RPKM , respectively . Fig 2B and S1 Table ) ., To validate the quantification of L1Hs autonomous transcription , we performed droplet digital PCR ( ddPCR ) to estimate autonomous and pervasive transcription levels on a reference panel of six cell lines: MCF-7 , K-562 , HeLa , HepG2 , SK-MEL-5 , and GM12878 ., For these experiments , we assumed that expression on the 5’ end of the L1Hs transcript can be used as an approximation to autonomous transcription due to the large imbalance of 5 truncated and full-length copies ., The expression on the 3’ end , on the other hand , is an approximation to the combination of autonomous and pervasive transcription ., We initially designed and tested multiple assays targeting different regions of the L1Hs locus and proceeded with the two best performing assays ( S2 Table ) ., The first assay targeted ORF1 , directly adjacent to the 5’UTR , representing the 5’ end of the transcript ., The second assay targeted ORF2 about 1 . 5 kb upstream of the 3’ UTR , representing the 3’ end of the transcript ., We completed the same design process for ORF2 to find the copy numbers of the truncated L1Hs transcripts ( i . e . , the transcripts missing the 5’ end of L1Hs ) ( Fig 2C , S3 Table ) ., Since autonomous transcription results in an enrichment full-length transcript of L1Hs , we estimated an approximation to the level of autonomous transcription as the pervasive transcription created by full-length transcription subtracted from the expression of the 5’ end ( ORF1 ) ., Fig 2D shows the relative quantification of L1Hs transcripts in these four cell lines using the HPRT1 5’ end as a reference ., We estimate that the ddPCR analysis detected approximately 2 , 412 . 8 copies of autonomously transcribed transcripts/ng in MCF-7 cells ., In agreement with our in-silico result , K562 and SK-MEL-5 had 990 . 0 and 1 , 075 . 6 copies of autonomously transcribed transcript/ng , respectively ., For the GM12878 cell line , we expected to find no autonomous expression of L1Hs; however , our ddPCR assays detected low levels of autonomous transcription of L1Hs ( 233 . 0 copies of autonomously transcribed transcript/ng; Fig 2D , S3 Table ) ., Overall , the quantification of L1Hs autonomous transcription using ddPCR was highly correlated with the quantification using TeXP ( Spearman correlation , rho = 0 . 9 , p-value = 0 . 01394 , S7 Fig ) ., This suggests that TeXP can remove most of the noise derived from pervasive transcription , although it is insensitive to samples with little LINE-1 autonomous transcription ( S8 Fig ) ., To address TeXP sensitivity in relation to noise we first tested TeXP under an ideal experimental setup ., In this simulation , the number of observed reads overlapping L1 subfamilies is a simple combination of known proportions of these signals ( i . e . pervasive and autonomous transcription ) ., For example , we simulate read counts where 30% of the reads derive from LIHs autonomous transcription and the remaining 70% derive from pervasive transcription ., We use simulated read counts as input to TeXP and calculate the root mean square error ( rmse ) between the known proportions and estimated proportions ., As observed in S15 Fig ( solid line ) , under this condition , the TeXP estimations of read counts is nearly identical to the simulated read counts ( median ( rmse ) = 0 . 0003 ) ., We also tested the effect of noise in the TeXP estimations ., To that end , we modeled the reads that cross-map across LINE-1 subfamilies as a noise component as a Poisson process deriving from random sequencing of cDNA fragments ., For these simulations , the Poisson noise was added to read counts deriving from of known proportions of signals as described above ., We tested two scenarios , one with noise equivalent to 10% of all observed reads ( dashed line ) and 20% of the of all observed reads ( long dashed lines ) ., In these conditions , the median TeXP RMSEs were respectively 0 . 014 and 0 . 028 ., Overall our observations suggest that , under low noise conditions , TeXP should be able to detect autonomous L1Hs transcription even when L1Hs has low transcription levels ., However , we also describe the TeXP capacity to detect low levels of L1Hs autonomous transcription degrades with higher levels of noise in RNA-se data ., In extreme situations , such as when 20% of the read counts are derived for a Poisson process , the RMSE can be up to 0 . 083 ., Researchers have long thought that LINE-1 instances are completely silenced in most healthy somatic cells ., LINE-1 is silenced by the methylation of its promoter 19 , which should preclude the transcription of mature LINE-1 mRNAs in healthy somatic tissue ., To test whether LINE-1 subfamilies are completely silenced in somatic tissue , we analyzed LINE-1 transcription in 7 , 429 primary tissue samples from the Genotype-Tissue Expression ( GTEx ) project 32 ( S4 Table ) ., Similar to the cell lines , we found that L1Hs was autonomously transcribed; L1P1 , L1PA2 , L1AP3 , and L1PA4 only had residual or spurious autonomous transcription in healthy tissues ( S9 Fig ) ., Furthermore , we found that pervasive transcription was the major signal in most RNA sequencing datasets , accounting for 91 . 7% , on average , of the reads overlapping LINE-1 instances ( S10 Fig and S14 Fig ) ., Overall , healthy tissues had a narrower range of L1Hs autonomous transcription levels than cell lines , with the peak transcription level of 47 RPKM ( Fig 2E ) versus 180 RPKM in the cell lines ( S1 Table ) ., We found no or very little ( <1 RPKM ) evidence of L1Hs autonomous transcription in 2 , 520 ( 34 . 3% ) of the GTEx RNA sequencing experiments from primary tissues ., Together , these results indicate that L1Hs is broadly transcribed in some healthy somatic tissues ., Therefore , if post-transcriptional regulatory constraints do not completely silence LINE-1 activity , one could expect that LINE-1 to play an important role in creating genetic diversity across somatic cells within an individual ., We then compared the landscape of LINE-1 subfamily transcription in Epstein-Barr virus ( EBV ) immortalized cell lines and their corresponding primary tissue to understand the changes induced by cell line immortalization ., EBV immortalization causes drastic changes in the expression of cell cycle , apoptosis , and alternative splicing pathways 34–36 ., Overall , we found that EBV-transformed cell lines derived from different tissues ( lymphoblastic and fibroblastic ) had distinct patterns of L1Hs autonomous transcription; lymphoblast ( blood-derived ) cell lines had no or little autonomous transcription of L1Hs ( S11 Fig ) with approximately 84% of samples having an estimated RPKM equal to zero , whereas fibroblastic ( skin-derived ) cell lines consistently had higher levels of L1Hs autonomous transcription ( median 1 . 5 RPKM ) with 58 . 7% of samples having an RPKM higher than 1 ., In general , EBV-immortalized cell lines reflected their tissue of origin ., While most ( 74 . 6% ) of the whole blood samples had no transcriptional activity of L1Hs , only one sample from skin had an L1Hs autonomous transcription level below 1 RPKM ., We further selected patients with both primary and EBV-transformed cell lines to assess whether the EBV transformation could change L1Hs autonomous transcription ., We found that both skin cells and lymphocytes had a drastic down-regulation of L1Hs autonomous transcription ( S12 Fig ) ., This finding suggests that EBV-transformed cell lines partially preserve the L1Hs transcription level from their tissue of origin , potentially explaining why fibroblast-derived induced pluripotent stem cells support higher levels of LINE-1 retrotransposition 37 ., Human solid tumors display increased levels of LINE-1 activity 38–40 ., In order to assess if this increase is a result from pervasive transcription or an increase in autonomous transcription in LINE-1 we run TeXP in RNA-seq from healthy thyroid samples and solid thyroid tumors ., We than calculate the distribution of number of reads deriving from pervasive and autonomous transcription ., We first observed that there indeed a significant difference in the number of reads from LINE-1 elements ( S13 Fig ) ., We also found that compared to healthy thyroid samples , tumor samples display higher levels of autonomous transcription and lower levels of pervasive transcription suggesting that autonomous transcription LINE-1 is driving most of the increase in LINE-1 expression in thyroid tumor samples ., Human tissues show remarkable variability of L1Hs autonomous transcription ., We found that L1Hs autonomous transcription is inversely correlated to the time it takes cells to divide ( cell turnover rate; Pearson correlation: cor = -0 . 6668968; p-value = 0 . 04 ) ., No correlation was found between cell turnover and pervasive transcription ( Pearson correlation: cor = 0 . 3983474; p-value = 0 . 2883 ) ., Tissues suggested to have low cell turnover , such as the human brain 41 , are amongst the tissues with the lowest levels of L1Hs autonomous transcription ( Fig 2E ) ., In particular , the human cerebellum , which has no transcription of L1Hs , is likely to have strong repression of L1Hs autonomous transcription ., This result seems to contradict the literature that suggests that the human brain supports high levels of somatic LINE-1 retrotransposition; however , most of these studies were based on neuro precursors that correspond to the early development stage of the human brain 15 , 42–44 ., Conversely , brain samples extracted from the striatum , putamen , and caudate , all regions associated with the basal ganglia , had higher levels of L1Hs autonomous transcription compared to other brain regions ( T-test basal ganglia vs . all other brain tissues , t = -7 . 0943; p value = 9 . 867e-12 –Fig 2E ) ; importantly , these levels were still low compared to other tissues ., Other tissues with low cell turnover rates , such as liver , pancreas , and spleen , also showed very little or no autonomous transcription of L1Hs ( 91 . 2% , 82 . 9% , 88 . 9% of samples , respectively , had a L1Hs RPKM < 1 –Fig 2E ) ., Conversely , germinative tissues have been proposed to support somatic activity of L1Hs elements 45 ., Our results suggest that this trend is more general , and most tissues associated with the reproductive system sustain higher levels of L1Hs autonomous transcription ( Fig 2E ) ., In addition , we found that the tissues with the highest levels of L1Hs autonomous transcription were enriched for high cell turnover; these included the nerve ( tibia ) , skin ( both exposed and not exposed to the sun ) , prostate , lung , and vagina ( Fig 2E ) ., Prior to this study , the effects of pervasive transcription on the estimates of transposable elements activity were largely ignored ., Here , we showed that most of the RNA-seq reads matching to LINE-1 instances derive from pervasive transcription , highlighting the importance of these effects ., In order to account for the effect of pervasive transcription on the quantification of LINE-1 activity we developed TeXP , a method that uses the widespread nature and the mappability signatures of LINE-1 subfamilies to account for and remove the effects of pervasive transcription from L1Hs , L1P1 , L1PA2 , L1PA3 , and L1PA4 subfamilies ., We compared TeXP estimates to other strategies such as naïve counts and others established methods such as SalmonTE 46 and TETranscript 47 ( S2 Fig ) ., Our estimations suggest that the pervasive transcription component is frequently missed and can be a confounding factor in some quantifications , for example , we observed that over 75% of reads mapping to ancient LINE-1 subfamilies derive from inactive subfamilies ( Fig 2A and S14 Fig ) ., We used TeXP to perform a comprehensive analysis of LINE-1 transcriptional activity across different cell types and somatic tissues ., Previous studies suggest that LINE-1 is active in germline and tumor cells , but not in most normal somatic cells with the exception of hints of activity in neuro-precursor cells 21 ., Somatic mosaicism of transposable elements , in particular LINE-1 , has been carefully characterized in the human brain and , despite some disagreement on the exact rate that LINE-1 retrotranspose in the human brain , it is clear that LINE-1 frequently create somatic copies in the human brain , in particular , in neuroprecursor , cortex and caudate nucleus cells 15 , 48 , 49 ., The somatic mobilization in other tissues , nonetheless , remain obscure and lack a systematic investigation ., A first step is to comprehensive characterize other tissues in terms of their LINE-1 transcriptome activity ., Surprisingly , we found that LINE-1 was active in many healthy human tissues , particularly in epithelial cells ., As we only detected a limited amount of LINE-1 activity in adult brain cells , our findings are in agreement with LINE-1 activity correlating with cell proliferation rate ., We validated many aspects of TeXP using ddPCR probes designed to quantify pervasive and autonomous transcription of L1Hs across human cell lines ., These results show that our method lacks the sensitivity to measure autonomous transcription ., In particular , TeXP underestimates L1Hs transcription levels when the signal-to-noise ratio is small ., One could imagine using unique regions of LINE-1 instances after removing the pervasive transcription signal to improve the transcriptional quantification ., However , removing pervasive transcription from individual instances is not trivial and should be carefully investigated ., As a tool , TeXP could be useful in several scopes beyond this study ., Pervasive transcription should also affect the quantification of other transposable elements and repetitive regions accounting for large portions of the human genome ., Our method could be further used to estimate the autonomous transcription levels of pseudogenes , SINEs ( ALUs ) and HERVs ., Furthermore , TeXP could be used in model organisms to distinguish the effects of pervasive transcription beyond humans ., The mouse genome , for example , has evidence of higher rates of retrotransposition but little is known about the activity of LINE-1 in somatic cells ., Moreover , some of the results we describe here could be extended and uncover important biological insights ., For example , assays such as induced pluripotent stem cells clones 50 , RC-seq 51 and L1-seq 52 , among others , could be used to carefully characterize of the rate of somatic retrotransposition in tissues with higher rates of autonomous transcription of L1Hs ., Additionally , TeXP could be used to further investigate the biases found in individuals from different ancestral backgrounds 53 ., The TeXP approach to quantify transcriptional activity by removing pervasive transcription could also be expanded to investigate the activity of LINE-1 during embryonic development or in pathological tissues , such as tumors ., TeXP models the number of reads overlapping L1 elements as the composition of signals deriving from pervasive transcription and L1 autonomous transcripts from distinct L1 subfamilies ., Our model proposes that the number of reads overlapping L1Hs instances as described by the Eq 1:, OL1Hs=T*GL1Hs*εpervasive+T*ML1Hs , L1Hs*εL1Hs+T*ML1Hs , L1PA2*εL1PA2+⋯+T*ML1Hs , j*εj, Where OL1Hs is the observed number of reads mapping to L1Hs , T is the total number of reads mapped to L1 instances , GL1Hs defines the proportion of L1 bases in the genome annotated as L1Hs , εpervasive is the percentage of reads emanating from pervasive transcription , M is the mappability fingerprint ( defined bellow ) that describes what is the proportion of reads emanating from the signal j ∈ {L1Hs , L1P1 , L1PA2 , L1PA3 , L1PA4} that maps to L1 subfamily i ∈ {L1Hs , L1P1 , L1PA2 , L1PA3 , L1PA4} and ε is the percentage of reads emanating from the autonomous transcription L1 Subfamily, j . This model can be further generalized as the Eq 2:, Oi=T ( Giεpervasive+∑jMi , jεj ), We selected these five LINE-1 subfamilies based on the rates of cross-mappability of simulated data ( S3 Fig ) ., In particular , we are interested in removing the effect of pervasive transcription from the estimates of L1Hs autonomous transcription ., We simulated reads from L1Hs transcripts we observed that most ( >90% ) of the reads emanating from L1Hs autonomous transcription map to the L1Hs , L1PA2 , L1PA3 , L1PA4 and L1P1 subfamilies ., Older subfamilies such as L1PA5 and L1PA6 , for example , correspond to less than approximately 5% of L1Hs cross-mappable reads ., In contrast to L1Hs which only a fourth of the reads map back to L1Hs , older elements such as L1PA5 and L1PA6 have a much higher self-mapping rates , 60% and 70% respectively ., Therefore , these subfamilies should be less affected by confounding factors deriving from pervasive transcription ., The number of reads mapped to each subfamily Oi is measured by analyzing paired-end or single-end RNA sequencing experiments independently ., TeXP extracts basic information from fastq raw files such as read length and quality encoding ., Fastq files are filtered to remove homopolymer reads and low quality reads using in-house scripts and FASTX suite ( http://hannonlab . cshl . edu/fastx_toolkit/ ) ., Reads are mapped to the reference genome ( hg38 ) using bowtie2 ( parameters:—sensitive-local -N1—no-unal ) ., Multiple mapping reads are assigned to one of the best alignments ., Reads overlapping LINE-1 elements from Repeat Masker annotation of hg38 are extracted and counted per subfamily ., The total number of reads T is defined as T = ∑iOi ., Pervasive transcription is defined as the transcription of regions well beyond the boundaries of known genes 27 ., We rationalized that the signal emanating from pervasive transcription would correlate to the number of bases annotated as each subfamily in the reference genome ( hg38 ) ., We used Repeat Masker to count the number of instances and number of bases in hg38 annotated as the subfamily i ∈ {L1Hs , L1PA2 , L1PA3 , L1PA4 , L1P1} ., We define Pi as the proportion of LINE-1 bases annotated as the subfamily i in the Eq 3:, Pi=Bi∑jBj , j∈{L1Hs , L1PA2 , L1PA3 , L1PA4 , L1P1}, On the other had mappability fingerprints , which represents how reads deriving from LINE-1 transcripts would be mapped to the genome , are created by aligning simulated reads deriving from putative L1 transcripts from each L1 subfamily ., For each L1 subfamily , we extract the sequences of instances based on RepeatMasker annotation and the reference genome ( hg38 ) ., Read from putative transcripts are generated using wgsim ( https://github . com/lh3/wgsim-parameters:-1 RNA-seq mean read length–N 100000 -d0 –r0 . 1 -e 0 ) ., One hundred simulations are performed and reads are aligned to the human reference genome ( hg38 ) using the same parameters described in the model session ., The three-dimensional count matrix C is defined as the number of reads mapped to the subfamily i ∈ {L1Hs , L1PA2 , L1PA3 , L1PA4 , L1P1} emanating from the set of full-length transcripts j ∈ {L1Hs , L1PA2 , L1PA3 , L1PA4 , L1P1} in the simulation, k . The matrix M is defined as the median percentage of counts across all simulations as in Eq 4:, Mi . j=mediank∈{1 , 2 , ., . , 100} ( Ci , j , k∑f∈{L1Hs , L1PA2 , L1PA3 , L1PA4 , L1P1}Ci , f , k ), We tested whether different aligners yield different mappability fingerprints ., BWA , STAR , and bowtie2 yielded similar results ( S15 Fig ) ., As L1 transcripts are not spliced , we decided to integrate bowtie2 as the main TeXP aligner ., We further tested the effect of read length on L1Hs subfamily mappability fingerprints ( S16 Fig ) ., To counter the effects of distinct read lengths TeXP constructs L1 mappability fingerprints libraries based on fastq read length ., We simulated reads emanating from their respective L1 subfamily transcripts and aligned these reads to the human reference genome creating a mappability fingerprint for each L1 subfamily ( S3 Fig ) ., When we analyzed the L1 subfamily mappability fingerprints we observed that younger L1 subfamilies tend to have more reads mapped to other L1 subfamilies ., For example , we find that only approximately 25% of reads from L1Hs ( the most recent–and supposedly active L1 ) maps back to loci annotated as L1Hs ., While older subfamilies such as L1PA4 , have a higher proportion of reads mapping back to its instances ( ~70%—S3 Fig ) ., The known variables Oi , T , the vector Pi , the mappability fingerprint matrix Mi . j are used to estimate the signal proportion ε and ϵ in Eq 2 by solving a linear regression ., We used lasso regression ( L1 regression ) to maintain sparsity ., We used the R package penalized ( 54—parameters: unpenalized = ~0 , lambda2 = 0 , positive = TRUE , standardize = TRUE , plot = FALSE , minsteps = 10000 , maxiter = 1000 ) ., TeXP was developed as a combination of bash , R and python scripts ., The source code is available at https://github . com/gersteinlab/texp ., A docker image is also available for users at dockerhub under fnavarro/texp ., Raw RNA sequencing datasets from healthy tissues were obtained from Database of Genotypes and Phenotypes ( DB-Gap - https://dbgap . ncbi . nlm . nih . gov ) accession number phs000424 . v6 . p1 ., Raw RNA sequencing data from cell lines were obtained from the ENCODE data portal ( https://www . encodeproject . org/search ) ., We selected RNA-seq experiments from immortalized cell lines with multiple cellular fractions and transcripts selection experiments ., Accessions and cell lines are available in S1 Table ., Mutation load and cell turn-over rate were extracted from the compilation of somatic mutation rate in Tomasetti et al 55 ., More ancient elements such as DNA transposons and LINE-2 have been shown to be primarily transcribed pervasively , hitchhiking the transcription of nearby autonomously transcribed regions 32 ., Therefore , we tested whether our estimation of L1Hs transcription level correlated with genes containing or adjacent to L1Hs instances ., We found no significant difference between the correlation distribution of a random set of genes and genes with L1Hs in exons or introns or within 3kb upstream or 3kb downstream of L1Hs ., This finding indicates that our estimation of L1Hs autonomous transcription is not significantly influenced by non-autonomous L1Hs transcription adjacent or contained by protein-coding genes’ loci ., Furthermore , we tested if and enrichment of pervasive transcription deriving from intronic regions would create a background signal distinct from the pervasive transcription derived from a whole genome model ., We correlated the number of LINE-1 instances from each subfamily in intergenic and intronic regions based on GENCODE v29 and found a statistically significant correlation between the number of instances in both regions ( Spearman corr = 0 . 979057 , p-value < 2 . 2e-16—S17 Fig ) ., All the cell lines used in this study were obtained from the American Type Culture Collection ( ATCC ) ( Manassas , VA , USA ) ., MCF-7 cells were cultured in Dulbecco’s Modified Eagle Medium: Nutrient Mixture F-12 ( DMEM/F12; Gibco ) ., HeLa , SK-MEL-5 , and HepG2 cells were cultured in Dulbecco’s Modified Eagle Medium ( DMEM; Gibco ) ., K562 and GM12878 cells were cultured in RPMI 1640 ( Gibco ) ., All cell culture media were supplemented with 10% fetal bovine serum ( FBS ) ( Atlanta Biologics ) and 1% penicillin/streptomycin ( Fisher Scientific ) ., All cells were cultured and expanded using the standard methods ., RNA was extracted using the RNeasy PLUS Mini Kit and the QIAshredders ( Qiagen ) following the manufacturer’s protocol ., All samples were treated with DNase I ( New England BioLabs Inc . ) to remove any remaining genomic DNA ., RNA concentration was determined by Qubit 2 . 0 Fluorometer ( Invitrogen ) ., RNA quality was determined by Nanodrop ( Thermo Scientific ) and 2100 BioAnalyzer with the Agilent RNA 6000 Nano kit ( Agilent Technologies ) ., Approximately 5 μg of RNA was used for synthesis of the cDNA using the iScript Advanced cDNA Synthesis Kit ( Bio-Rad ) ., The final cDNA product was quantified and a working solution of 10 ng/μL was prepared for the subsequent studies ., Droplet Digital PCR ( ddPCR ) System ( Bio-Rad Laboratories ) was utilized to quantify the L1Hs transcript expression in the cell lines described above ., Since L1Hs is a highly repetitive and heterogeneous target , we had initially designed and tested a panel of primers and probes that targeted the 5’ untranslated region ( 5’UTR ) , the open reading f | Introduction, Results, Discussion, Materials and methods | The Long interspersed nuclear element 1 ( LINE-1 ) is a primary source of genetic variation in humans and other mammals ., Despite its importance , LINE-1 activity remains difficult to study because of its highly repetitive nature ., Here , we developed and validated a method called TeXP to gauge LINE-1 activity accurately ., TeXP builds mappability signatures from LINE-1 subfamilies to deconvolve the effect of pervasive transcription from autonomous LINE-1 activity ., In particular , it apportions the multiple reads aligned to the many LINE-1 instances in the genome into these two categories ., Using our method , we evaluated well-established cell lines , cell-line compartments and healthy tissues and found that the vast majority ( 91 . 7% ) of transcriptome reads overlapping LINE-1 derive from pervasive transcription ., We validated TeXP by independently estimating the levels of LINE-1 autonomous transcription using ddPCR , finding high concordance ., Next , we applied our method to comprehensively measure LINE-1 activity across healthy somatic cells , while backing out the effect of pervasive transcription ., Unexpectedly , we found that LINE-1 activity is present in many normal somatic cells ., This finding contrasts with earlier studies showing that LINE-1 has limited activity in healthy somatic tissues , except for neuroprogenitor cells ., Interestingly , we found that the amount of LINE-1 activity was associated with the with the amount of cell turnover , with tissues with low cell turnover rates ( e . g . the adult central nervous system ) showing lower LINE-1 activity ., Altogether , our results show how accounting for pervasive transcription is critical to accurately quantify the activity of highly repetitive regions of the human genome . | Repetitive sequences , such as LINEs , comprise more than half of the human genome ., Due to their repetitive nature , LINEs are hard to grasp ., In particular , we find that pervasive transcription is a major confounding factor in transcriptome data ., We observe that , on average , more than 90% of LINE signal derives from pervasive transcription ., To investigate this issue , we developed and validated a new method called TeXP ., TeXP accounts and removes the effects of pervasive transcription when quantifying LINE activity ., Our method uses the broad distribution of LINEs to estimate the effects of pervasive transcription ., Using TeXP , we processed thousands of transcriptome datasets to uniformly , and unbiasedly measure LINE-1 activity across healthy somatic cells ., By removing the pervasive transcription component , we find that ( 1 ) LINE-1 is broadly expressed in healthy somatic tissues; ( 2 ) Adult brain show small levels of LINE transcription and; ( 3 ) LINE-1 transcription level is correlated with tissue cell turnover ., Our method thus offers insights into how repetitive sequences and influenced by pervasive transcription ., Moreover , we uncover the activity of LINE-1 in somatic tissues at an unmatched scale . | sequencing techniques, genetic fingerprinting, cultured fibroblasts, biological cultures, human genomics, dna transcription, genetic elements, molecular biology techniques, rna sequencing, research and analysis methods, genomic signal processing, artificial gene amplification and extension, gene expression, cell lines, molecular biology, signal transduction, genetic fingerprinting and footprinting, cell biology, genetics, transposable elements, biology and life sciences, genomics, mobile genetic elements, cell signaling, polymerase chain reaction | null |
journal.pcbi.1000785 | 2,010 | A Model for Genetic and Epigenetic Regulatory Networks Identifies Rare Pathways for Transcription Factor Induced Pluripotency | Cellular states are plastic , and even terminally differentiated cells ( e . g . , B-cells ) can be reprogrammed to pluripotency by ectopic expression of selected transcription factors 1 , 2 , 3 , 4 , 5 , 6 , 7 ., This finding raises the possibility of creating patient-specific stem cells for regenerative medicine 8 ., However , reprogramming efficiencies range from 0 . 0001% to 29% 5 , 6 , 9 , 10 , with most reports showing that successful induction of the pluripotent state is rare even if all required factors are present 11 , 12 ., The genetic and epigenetic regulatory mechanisms that make reprogramming possible , and determine its efficiency , are poorly understood 2 ., Elucidating these mechanistic principles can help define optimal strategies for reprogramming differentiated cells , and answer fundamental questions regarding how cellular identity is maintained and transformed ., In spite of recent progress , our knowledge of the identities and functions of the genes and proteins involved in regulating the transformation of cellular identity is grossly incomplete 2 , 13 , 14 ., Thus , it is not yet possible to construct a detailed molecular mechanistic description of how epigenetic modifications and expression of master regulatory genes are controlled ., However , ectopic expression of the same transcription factors can reprogram different cell types 1 , 6 , 12 , and the genetic and epigenetic transformations observed during reprogramming of diverse differentiated cells share many common features 2 , 11 , 15 , 16 , 17 , 18 , 19 ., These common observations can be the basis for developing a conceptual understanding of the general architecture of the genetic and epigenetic networks that regulate transcription factor induced reprogramming and establish cellular identity during differentiation ., We have taken a step toward this goal by developing a computational model that is consistent with , and suggests general mechanistic explanations for , empirical observations of transcription factor induced reprogramming ., The model makes experimentally-testable predictions ., If validated , descendents of this model could also provide insights into the aberrant de-differentiation events which characterize some of the most malignant cancers ., Elegant theoretical models for the molecular regulatory networks responsible for stem cell renewal and differentiation and the population dynamics of these processes have been created 20 , 21 , 22 , 23 , 24 ., Our goal is different ., We aim to develop a model for the architecture of coupled epigenetic and genetic networks which describes large changes in cellular identity ( e . g . , induction of pluripotency by reprogramming factors ) ., Although the general principles of interactions between genetic and epigenetic layers of regulation have been described 25 , 26 , no computational model has been developed to study the outcomes of such interactions and their biological consequences ., Such a computational model would be a useful complement to experiments in understanding the processes that occur during reprogramming of differentiated cells , and why reprogramming is rare ., Here , we propose , to our knowledge , the first computational model that describes how cellular identity changes by creating a mathematical description of interactions between epigenetic and genetic networks ., Our goal is not to describe the details of how specific regulatory proteins interact , but rather , to understand general principles underlying how cellular states evolve upon ectopic expression of certain types of genes ., The concise model we have developed explains why reprogramming probability is low , and makes experimentally testable predictions ., Almost all cells in a multi-cellular organism share the same DNA sequence ., Yet , different cell types express distinct genes and perform different functions ., Epigenetic modifications are major regulators of cell-type specific gene expression ., They function by packaging DNA into configurations that allow only some genes to be expressed , while other genes are tightly packed into heterochromatin structures that hinder access of most transcription factors 27 ., Changes in cellular identity during developmental differentiation or transcription factor induced reprogramming require modification of the epigenetic state of the cell ., The maintenance and alteration of cellular identity is regulated by a complex set of interactions between developmentally important genes , chromatin modifiers , transcription factors etc . , the details of which remain unknown ., Toward developing a model for the architecture of these complex regulatory networks we consider only the developmentally important genes ., For simplicity , each ensemble of genes responsible for maintenance of a particular cellular identity ( e . g . , Oct4 , Sox2 , etc . , for pluripotency ) is described as a single module ( Fig . 1a ) ., Theoretical justification for treating genes that control the embryonic stem ( ES ) cell state as a collective unit exists 28 ., We also carried out some studies with each module consisting of a small number of genes ( see corresponding discussion below ) ., ES cells can differentiate into various lineages ., Upon further differentiation , cells become more restricted ., For example , hematopoetic stem cells can differentiate into T and B-lymphocytes , but not neural cells ., Therefore , in our model , we arrange gene modules in a hierarchy ( Fig . 1a ) ., Although each cell state can potentially differentiate into many branches , without loss of generality , we consider two branches to emanate from each cell state ., Thus , the cellular states are arranged on a Cayley tree ., In our model , a cell state ( Fig . 1b ) is specified by: i the state of the epigenome , and ii the expression levels of master regulatory genes ., ES cells are cultured in specific media ( e . g . , containing LIF/BMP4 for mouse ES cells ) to prevent differentiation 43 ., The medium inhibits a self-induced differentiation pathway ., We represent this feature by assuming that proteins expressed by the module regulating the ES state favor putting positive chromatin marks on gene modules regulating immediate progenies if LIF , etc . are absent ., Simulations of this situation show ( Fig . 4 ) that , as in experiments 2 , ES cells differentiate randomly to one of their progeny ., Our model exhibits robust differentiation ( forward programming ) to specific cell states when the appropriate cues are delivered ., Appropriate cues are expression of proteins ( e . g . , signaling products ) that become available during interphase ., In the next telophase , these proteins favor putting positive histone marks on the gene module regulating the appropriate progeny of the current cellular state ( rule 1 ) ., Results from our computer simulations ( Fig . 4 in Text S1 ) demonstrate that our model exhibits high-fidelity responses to such differentiation cues ., This is consistent with the experimental observation that overexpression of the master-regulatory genes of desired lineage leads to predominant differentiation in that direction 44 , 45 ., This result is relevant because practical use of induced pluripotent cells will involve differentiating them to desired cell types ., We also find an exponential decay of the number of progenitor cells ( with a signal strength-dependent lifetime ) , as has been noted before 46 ., We simulate reprogramming experiments by starting with a terminally differentiated cell state where genes from other lineages , etc . , have been epigenetically silenced ., Our basic premise is that terminally differentiated cells can reprogram because protein products of the ectopically expressed genes can potentially alter the epigenetic state of the cell as a cell progresses through the telophase ., In our low resolution model , we identify genes not by names , but rather by their functional properties ., We presume that Klf4 and c-Myc are important ingredients of the reprogramming “cocktail” because they promote progression through the cell cycle , and this provides more opportunities for the other reprogramming factors to perturb the epigenome during telophase ., This functional identification of Klf4 and c-Myc makes our model general , and is validated by experiments showing that shutting down p53 abrogates the need for Klf4 and c-Myc for reprogramming ( only Oct4 and Sox2 required ) precisely because this also allows faster progression through the cell cycle 47 , 48 , 49 , 50 , 51 ., ( Interestingly , simulataneous action of c-Myc and p53 knock-down decreases the efficiency of reprogramming indicating existence of the optimum ) ., Oct4 and Sox2 have an enormous number of binding targets on the DNA , and are responsible for maintenance of the ES state which likely implies multiple interactions with master-regulatory genes ., We therefore identify the ectopic expression of these genes with the function of being highly likely to perturb the epigenome during telophase ., Each gene module in our model corresponds to an ensemble of carefully tuned mutually interacting master-regulatory genes that govern a particular cellular identity ., At the moment , not all of the master-regulatory genes of cellular states are experimentally identified , thus we use gene modules to represent these ensembles in a general way ., Even though products of ectopically expressed Oct4 and Sox2 have numerous targets 52 , it is unlikely that the epigenetic state of many such sets of genes will be simultaneously altered ., Thus , in order to mimic the effect of reprogramming factors , we randomly pick one epigenetically silenced gene module and change its state to correspond to open chromatin ., To examine the effects of overexpression of ectopic genes , we also study the consequences of multiple epigenetic transformations at a time ( see discussions below ) ., Starting with a terminally differentiated state we perturb the epigenome as described above , and then simulate the next gene expression phase where both the module regulating the terminally differentiated state and the one which was transformed to open chromatin status can express proteins according to rules 1′–2′ ( or Eq . 3 ) ., The protein atmosphere thus generated becomes the input to simulation of the next telophase according to rules 1–4 ( or Eq . 2 ) ., This can then potentially establish a new epigenetic state which becomes input to simulation of the next gene expression phase; i . e . , the genetic and epigenetic states are allowed to come to a new balance ., Then , the epigenetic state of another randomly picked silent gene module is changed to open chromatin because of the effects of reprogramming factors ., This procedure is continued until a fully reprogrammed or a dead/arrested state is achieved ( see below ) ., We carried out 10 , 000 independent replicate simulations of the effects of ectopic expression of reprogramming factors on a differentiated cell in a model with four levels in the hierarchy of cellular states ., Results from each simulation describe the fate of a single cell in a population ., Only 3 out of 10 , 000 “cells” successfully reprogrammed; i . e , as in experiments , reprogramming is rare ., The percentage of cells that reprogram depends upon the number of levels in the hierarchy ( 0 . 0001% and 2% of the cells reprogram successfully for a five-level and three-level hierarchy , respectively ) ., This suggests that reprogramming efficiency should improve for less differentiated cells ., This has been demonstrated directly in a well-defined lineage such as the hematopoietic system 53 ., Additionally , Hanna et al . demonstrated a notable increase in the efficiency of reprogramming B cells upon Pax5 knockdown 12 ., Loss of Pax5 had been previously shown to cause dedifferentiation of B cells to a common progenitor that upon transplantation allowed T cell development 54 ., We report results for models consisting of 3- , 4- and 5-levels in the hierarchy of gene modules , but in real organisms the depth of the differentiation tree could be as large as tens of levels 55 ., Since our results indicate that reprogramming efficiency decreases quickly with the increase in the depth of the hierarchy , it is natural to ask why reprogramming is at all feasible ., The reason is that master-regulatory genes that regulate closely related states are not mutually exclusive sets of genes ., The difference between genes that regulate closely related cellular states can be as small as one or two genes 54 ., However , genes that regulate cellular states distal in the hierarchy are not correlated in this way ., As our model does not treat correlations between genes that regulate closely related states , in effect , each gene module in our model represents master regulatory genes that control the identity of a number of cellular states that have many master regulatory genes in common ., Thus , a 5-level hierarchy in our model might represent a 50-level depth of differentiation in a real organism ., The results reported above were obtained for specific values of parameters ( Table, 2 ) which represent rules 1–4 and 1′–2′ ( Eqs . , 2–3 in Methods ) ., Our simulation results are consistent with diverse experimental observations ( see Table 3 and discussion below ) only if the methylation constraints ( rule 4 ) and mutual repression of expression of gene modules ( rule 2′ ) are relatively strong effects ( i . e . H>G and J>F , see Table 2 , Eqns ( 2–3 ) , and parameter sensitivity in SI for further details ) ., As long as these two conditions are met , the specific choice of parameter values only alters the quantitative value of the number of successfully reprogrammed cells , but reprogramming to the ES state remains rare ., Our simulation results suggest a mechanistic explanation for why reprogramming is so rare ., When reprogramming factors attempt to change cellular identity by altering the epigenetic state of a previously silenced gene module , the probability of success depends upon the position of this module relative to the one that regulates the terminally differentiated state ., We find that the position of the module whose epigenetic state is altered can belong to one of three categories ( Fig . 5a ) ., Suppose this gene module regulates a cellular identity in a different lineage from the terminally differentiated state ., In the next interphase , both modules can express proteins as there are no mutually repressive interactions between them ., In the subsequent telophase , proteins expressed by each module would favor epigenetic silencing of the other ( rule 4 ) ., Expression of proteins characteristic of a cell type from a different lineage does not favor reprogramming because it leads to cell death or arrest in our model ., Cell death could be mediated by various mechanisms including genetic instabilities if the two open gene modules send conflicting instructions to housekeeping genes ., Of course , there is also the chance that the cell will be rescued by stochastic expression of some master-regulatory gene , or that the cell will assume an “intermediate” cell state without master regulation that could be viable , but does not reprogram , such as some arrested states 18; finally , there is a possibility that two master regulators will not repress each other in full , but some minuscule amount of expression of both will remain thus , arresting the cell ., Within the framework of our model we do not distinguish between these possibilities , and classify cells in all these unusual , dead , or arrested states to be dead/arrested ., The gene module whose epigenetic state is altered by reprogramming factors could be in the same lineage as the differentiated cell , but not be its sibling or progenitor ., In the following interphase , this module and the one that regulates the terminally differentiated state can both express proteins ., In the subsequent telophase , according to our model , protein products of the gene module regulating the terminally differentiated state will favor epigenetic silencing of the module that was turned on by the action of reprogramming factors ( rule 4 ) ., But , the opposite is not true because the cellular state regulated by the gene module whose epigenetic state was altered by reprogramming factors could potentially differentiate to the terminally differentiated cell type ., Thus , the altered gene module will be silenced again , and the cell remains terminally differentiated ., Reprogramming factors could also change the epigenetic state of a previously silenced gene module which regulates an immediate sibling or the progenitor of the terminally differentiated state ., In the subsequent interphase , these two gene modules with open chromatin status will not simultaneously express proteins at high levels ., This is because gene modules that are “nearest neighbors” in the hierarchy mutually repress each other ( rule 2′ ) ., If the dominantly expressed gene module ( determined stochastically ) is the one which regulates a sibling or the progenitor of the terminally differentiated state , then during the next telophase its products will establish epigenetic marks consistent with a new identity ( rule 1 ) ., Thus , with a probability determined by stochastic effects , a step toward reprogramming can occur via trans-differentiation or de-differentiation ., These arguments suggest that a step toward reprogramming occurs with significant probability only if the epigenetic state of a gene module regulating a sibling or progenitor of the differentiated cell is changed to open chromatin status by reprogramming factors ., This is a rare event in our simulations where the set of master regulator genes that determine a cellular identity are considered to be one gene module ., In reality , this is even less likely because it requires reprogramming factors to orchestrate changes to a set of master regulator genes synchronously ., For successful reprogramming to the ES state , a sequence of such rare events must occur in a particular cell ., This is because after a step toward reprogramming occurs , the partially reprogrammed cell is subject to all the constraints discussed above ., Therefore , although cellular identity is plastic , reprogramming a terminally differentiated cell to the ES state is rare and requires many cell cycles ., Two examples of how states evolve under the influence of reprogramming factors in our simulations are shown in Fig . 5b ., The first example shows a “cell” that does not successfully reprogram , as after a successful trans-differentiation , ultimately the cell is arrested/dead ., In the second example reprogramming to the ES state occurs successfully , and it shows an interesting feature ., At an intermediate time point , before the ES state is realized , reprogramming factors have turned on expression of the endogenous gene module that regulates the ES state ., But this is transient , as this module is quickly silenced ., We find that , unless proteins expressed by each gene module can stably repress genes that are distal in the hierarchy of states ( rule 4 , realized presumably through DNA methylation ) , expression of endogenous genes that regulate the ES state can occur early and prior to the temporal increase in the number of bivalently marked genes observed during reprogramming ., In other words , our model recapitulates the observation that endogenous expression of Oct4 and Sox2 is the last step toward reprogramming only if the “DNA methylation” constraint is long-ranged ., Thus , the model suggests that transient blocking of methylation machinery might allow endogenous expression of Oct 4 , Sox2 , etc . , at intermediate time points ., This is consistent with the observation that DNA methyltransferase and histone deacetylase ( HDAC ) inhibitors , such as valproic acid ( VPA ) , an HDAC inhibitor , improve reprogramming efficiency 9 ., Our model predicts that reprogramming occurs via a sequence of trans-differentiations to immediate siblings or de-differentiations to immediate progenitors in the hierarchy of cellular states ., Note , however , that our results do not imply that pure differentiated states will be observed as reprogramming occurs ., Oct4 , Sox2 , etc . , have numerous targets , and so genes from unrelated lineages will transiently be expressed during reprogramming to the ES state ( 22 ) ., But , the entire set of master regulatory genes for a cellular state from a different lineage will not be expressed ., We illustrate this point by showing computer simulation results from a model where we consider each gene module to be comprised of three individual genes ( Fig . 6 ) ., Reprogramming factors can attempt to change the epigenetic state of the individual genes randomly as before ., However , in this more complex model , if we allow only one genes epigenetic state to be modified in every telophase , reprogramming becomes so rare that we cannot observe it in a realistic computer simulation time ., So , we allowed a larger number of transformations per cycle ., Choosing this number to be too large corresponds to overexpression of reprogramming factors , and this severely hinders reprogramming ( Text S1 , section 2 ) ., For the results shown in Fig . 6 , we randomly pick 12 genes and change their epigenetic states during each simulated telophase ., We assume that the entire set of genes comprising a module must be expressed for its products to regulate the epigenetic or genetic network ., This is consistent with combinatorial control of regulation ., Fig . 6a shows two examples of in silico cells that successfully reprogram to the ES state ., Reprogramming takes place via a sequence of trans-differentiation and de-differentiation events wherein the entire set of genes that regulate a progenitor or sibling of the previous cellular state is expressed ., But , the intermediate states are not pure differentiated states as some genes from unrelated lineages are also turned on at the same time ( as observed in experiments 18 ) ., If the terminally differentiated state in our simulations is analogous to a B cell , our simulations predict that all successfully reprogrammed cells must transit through an impure state where all the genes regulating the hematopoetic stem cell state are turned on ( as in Fig . 6a ) ., Although beyond the scope of this work , it would be reasonable to test this prediction by applying a cre-lox based lineage-tracing approach ., Using one or more stem/progenitor specific promoters that are inactive in the terminal state ( e . g . , B cell ) , in combination with a lox-STOP-lox reporter , one could retrospectively determine whether all the resulting iPS cells are labeled and hence have transiently expressed markers of earlier stages within the same lineage ., An unrelated cell type , such as fibroblasts , should generate unlabeled iPS cells because it would not be expected to transition through hematopoietic progenitor stages and hence serve as an appropriate control ., The results depicted in Fig . 6 could also potentially be assessed quantitatively in experiments where the temporal evolution of the gene expression patterns of a number of successfully reprogrammed cells is observed ., Consider a state where the master regulator genes corresponding to a particular cellular identity are all expressed ., One could then ask: when these genes are subsequently silenced during reprogramming , which complete set of master regulatory genes start expressing proteins ?, One could ask this question at various times during reprogramming and in various successfully reprogrammed cells ., This would enable calculation of the following four point correlation function ( C ) : ( 1 ) where δ is the Kroenecker delta , t is time , t+Δt is a later instant in time during reprogramming ( a cycle in our simulations ) , i and j are labels of two genes , and Si is either 1 or 0 depending upon whether the ith gene is expressing proteins or turned off ., Our computer simulations predict ( Fig . 6b ) that , at each stage of reprogramming , the correlation function would have high values for genes from lineages related to the terminally differentiated starting point and low values for genes of unrelated lineages ., We hope that this prediction can also be assessed in future experiments ., This could involve permanent labeling as mentioned above , or possibly , in the long-term , real-time monitoring of cell state transitions ., To the best of our knowledge , we have developed the first computational model that describes how terminally differentiated cells may be reprogrammed by expression of ectopic genes ., This is achieved by a mathematical description of interactions between epigenetic and genetic networks of master-regulatory genes that govern specific cell states ., The model also describes differentiation in accord with experiments ., Our model describes cellular states as attractors on a generalized landscape of all possible genetic/epigenetic configurations ., Cellular states are stable , self-renewing states unless a perturbing signal ( either differentiation cue or reprogramming factors are introduced ) ., As summarized in the table 3 , major features of the reprogramming process are explained by our results and the mechanism of reprogramming it suggests ., For instance , different cell types can be reprogrammed with the help of the same set of factors 1 , 12 , 16 because ectopic expression of genes that have many targets ( e . g . , Oct4 and Sox2 ) can perturb the epigenetic state regardless of the identity of the starting differentiated cell type ., The importance of fast progression through the cell cycle ( due to cMyc , Klf4 , or p53 knockdown ) is because this offers more opportunities for epigenetic transformations during telophase ., The important experimental observation that endogenous Oct4 and Nanog expression 2 occurs just prior to complete reprogramming is also recapitulated by our model ., The stochastic nature of the reprogramming process 56 and its low yield 2 are because only a few types of trajectories can lead to successful reprogramming , and they are realized rarely by stochastic perturbation of the epigenome by the reprogramming factors ., Our model predicts the nature of these rare trajectories to be those that progress through reprogramming via de-differentiation to closely related cell types ( immediate progenitors or siblings in the hierarchy ) ., Ways to directly test this prediction are suggested ., However , any feature that involves a specific molecular interaction between specific molecules is not described by our model ., In our current model , we consider states with genes that express proteins with conflicting demands to die/arrest ., In reality , some of these situations can give rise to steady states that do not arrest or reprogram ( such as the recently studied BIV1 , MCV8 , etc . , cell lines ) 18 ., The ideas emerging from our model are consistent with observations made by manipulating these trapped states ., For example , consider the observation that removing reprogramming factors allows cells from the BIV1 cell line ( isolated during reprogramming of B lymphocytes ) 18 to reprogram to the ES state ., This suggests that overexpression of reprogramming factors prevents these cells from reprogramming to the ES state ., Our model suggests that this could be due to two reasons ., First , over expression of reprogramming factors ( which have many targets ) could simultaneously change the epigenetic states of a number of silenced genes to permissive chromatin status ., Our simulations of the model shown in Fig . 6 with a large number of such simultaneous transformations ( e . g . , 22 at a time , rather than 12 at a time used for Fig . 5 ) prevents successful reprogramming because of the large probability of obtaining dead or arrested states ., As noted above , one of these states that cannot reprogram could correspond to the BIV1 cells ., Secondly , our model describes how lowering expression of reprogramming factors in BIV1 cells could enable reprogramming ., In our simulations , we consider proteins expressed during each interphase to act on the epigenome to reach a new balance which then leads to a corresponding protein expression pattern before another epigenetic transformation can occur due to the action of reprogramming factors ., This is analogous to assuming that the reprogramming factors can act to change the epigenetic state of a set of master regulator genes rarely ., If reprogramming factors are grossly overexpressed , this would not be true ., So , before a new protein expression pattern could be expressed consistent with a newly acquired epigenome ( say , de-differentiation to a progenitor ) , another epigenetic transformation would occur , and the whole cycle would start again ., Simulation results showing this effect upon overexpression of reprogramming factors are depicted in Fig . 3B in Text S1 ., Removing reprogramming factors could potentially allow reprogramming of cells trapped in such an infinite loop ., Our low-resolution model for the architecture of genetic and epigenetic regulatory networks that determine how cellular identities change is consistent with diverse observations ( Table 3 ) ., In formulating this model , we ruled out many models that were inconsistent with known experimental results , but we cannot rule out all other possible models ., Therefore , the predictions of the model ( noted earlier ) need to be experimentally tested ( perhaps in ways that we have suggested ) to either falsify it or encourage studying it further ., If tested positively , the suggestions emerging from our model regarding ways to enhance reprogramming yields should be further explored ., It would also be interesting to study other transcription factor induced cell state conversions 57 , 58 within the conceptual and computational framework we have developed for how cellular identity is transformed ., In particular , recent results of direct conversion between exocrine and endocrine cells through ectopic expression of three alternative transcription factors 59 should be examined ., It would be interesting to further investigate several assumptions adopted in the model for the lack of specific information about individual master-regulatory modules ., For example , maximum expression levels of different master-proteins within different modules could differ , as well as coupling between genetic and epigenetic networks could be different for different modules ., Also , we assumed that every simulated cell ( as represented by a simulated trajectory ) has the same level of expression of reprogramming factors while in reality cells can be transfected in a heterogeneous fashion ., Also , the difference in viral integration sites in different cells could lead to the different expression levels of exogeneous genes thus making effect of reprogramming factors heterogeneous across the population ., In a sense then , we have studied those cells which have expressed reprogramming factors at levels above a threshold ., It would be interesting to further explore the consequences of such heterogeneity ., Another avenue for further exploration lies in defining the notion of time during the reprogramming process , in this work cell cycling has been adopted as a measure of time required for reprogramming while in reality cells cycle with non-equal rates determined from some form of cell division rate distribution ( simplest form would be an exponential distribution ) ., It would be interesting to see applicability of the 4-point correlation function based analysis for the situation when cell cycling rates are not identical ., Finally , de-silencing action of reprogramming factors is assumed to be distributed randomly ., It would be interesting to consider situations when de-silencing distribution is not uniform across the hierarchy ., It is possible that non-uniform distributions can improve the reprogramming efficiency ., From the standpoint of statisti | Introduction, Results, Discussion, Methods | With relatively low efficiency , differentiated cells can be reprogrammed to a pluripotent state by ectopic expression of a few transcription factors ., An understanding of the mechanisms that underlie data emerging from such experiments can help design optimal strategies for creating pluripotent cells for patient-specific regenerative medicine ., We have developed a computational model for the architecture of the epigenetic and genetic regulatory networks which describes transformations resulting from expression of reprogramming factors ., Importantly , our studies identify the rare temporal pathways that result in induced pluripotent cells ., Further experimental tests of predictions emerging from our model should lead to fundamental advances in our understanding of how cellular identity is maintained and transformed . | Most cells in an organism have the same DNA ., Yet , different cell types express different proteins and carry out different functions ., These differences are reflected by cell epigenetics; i . e . , DNA in different cell types is packaged distinctly , making it hard to express certain genes while facilitating the expression of others ., During development , upon receipt of appropriate cues , pluripotent embryonic stem cells differentiate into diverse cell types that make up the organism ( e . g . , a human ) ., There has long been an effort to make this process go backward— i . e . , reprogram a differentiated cell ( e . g . , a skin cell ) to pluripotent status ., Recently , this has been achieved by overexpressing specific transcription factors in differentiated cells ., This method does not use embryonic material and promises the development of patient-specific regenerative medicine ., The mechanisms that make reprogramming rare , or even possible , are poorly understood ., We have developed the first computational model of transcription factor-induced reprogramming ., Results obtained from the model are consistent with diverse observations , and identify the rare pathways that allow reprogramming to occur ., If validated by further experiments , our model could be further developed to design optimal strategies for reprogramming and shed light on basic questions in biology . | biophysics/theory and simulation | null |
journal.pbio.0050239 | 2,007 | Heritable Stochastic Switching Revealed by Single-Cell Genealogy | Inheritance is more than the faithful copying and partitioning of genomic information ., When cells divide , the mother cell passes numerous other cellular components to the freshly born daughter , including nucleosomes , transcription factors , mitochondria , and substantial fractions of its proteome and transcriptome ., In this way , an entire pattern of gene expression can be passed from mother to daughter , a phenomenon known as epigenetic or non-Mendelian inheritance ., Classic examples permeate the literature and include the sex-ratio disorder in Drosophila 1 , the yellow-tip phenotype in melons 2 , the telomere position effect in yeast 3 and mouse 4 , and prions such as Psi+ in yeast 5 ., The time scale over which epigenetic phenotypes may persist spans many orders of magnitude and depends strongly on the physical mechanism used by the cell 6 ., In general , however , epigenetic phenotypes are substantially less stable than chromosomally inherited ones are 6 , 7 , and can change reversibly in single cells 3 , 8 , 9 during development 10 , 11 , or in mature organisms 12 ., Beginning with landmark studies on the lac operon in the 1950s , positive transcriptional feedback loops have emerged as a means to store cellular memory 13–15 ., Such epigenetic inheritance systems are frequently described as “bistable , ” meaning that transcriptional activity of genes in the network tends to become fixed in single cells around one of two stable levels ( ON and OFF ) , each of which is able to stably persist for many generations 8 , 16 , 17 ., Stochastic fluctuations in the creation or decay of the proteins involved 18–34 , or changes in external cues ( e . g . , a changing environment ) , are responsible for causing transitions between the two states 8 , 13 , 16 , 17 ., This flexible strategy , which is present in both prokaryotes and eukaryotes , allows genetically identical cells to diversify their population , possibly allowing them to exploit new environmental niches or to survive in a fluctuating external environment 35 ., Feedback-based cellular memories show an exceptional range of stability; depending on the strength of the feedbacks , cells may display memory of a previous expression state as short as a single generation to as far back as many thousands of generations 17 ., However , quantitative measurements of phenotype stability , switching , and heritability are rare , both because detailed genealogical relationships are challenging to produce in single cells 36 and because reporters indicating degree of inheritance are not always available ., To measure how a dynamic gene expression state is inherited , we focused on an engineered version of the galactose utilization ( GAL ) pathway in the yeast Saccharomyces cerevisiae ( Text S1 ) ., We disrupted the pathways major negative feedback loop and grew cells in conditions where only a single positive-feedback loop was operational ( see Materials and Methods ) ., Under these conditions , cells stochastically transition between two distinct expression states even in the absence of an extracellular trigger ., These infrequent switching events therefore likely arise from fluctuations in concentrations of regulatory proteins within the individual cells 37 ., We are able to monitor transitions between ON and OFF using a fluorescent reporter ( see Materials and Methods , Figure S3 ) ., Together , these attributes make our network an ideal model system that is well suited to study the heritability of an entire dynamic gene expression state ., In this work , we find that not only is the epigenetic phenotype itself heritable , but that the stability of this phenotype is likewise a heritable quantity ., In other words , when cells divide , the nascent daughter cell assumes both the expression state of the mother cell as well as its tendency to switch epigenetic states at a similar time in the future ., This is surprising , especially considering that individual cells viewed outside their genealogical context appear to switch completely at random ., We resolve this apparent contradiction using a simple stochastic model ., We first set out to quantify , using fluorescence microscopy , the infrequent switching events that occur at random times ., All experiments began with a single cell confined between a cover slip and a thick agar pad ., Over a period of about 920 min ( >15 h ) each cell grew and divided to eventually form a small colony of 50–100 cells ., Throughout the measurement period , these cells diverged in behavior , with some increasing in fluorescence and others decreasing ., We repeated this process with more than 100 progenitor cells , so in sum our data represent many thousand single-cell trajectories ., We present two examples of the experimental procedure in Figure 1 ., In Figure 1A , an initially bright cell develops into a small colony with distinct subpopulations ., The dim cells in the lower subpopulation continue to diminish in fluorescence with each successive cell division as the remaining molecules of green fluorescent protein ( GFP ) dilute ., In Figure 1B , an initially faint cell likewise gives rise to a variegated colony with cells both dim and bright ., Together , these two processes generate a broad bimodal steady-state distribution ., Narrowing our focus to initially OFF progenitor cells , we allowed each to grow , divide , and give birth to other initially OFF cells ., We then recorded instances when cells switched into the ON state ( Figure 2A and Video S1 ) ., Because cellular auto-fluorescence is uniformly small throughout the population of OFF cells , these fluorescing events were generally distinguished unambiguously from background fluctuations ., Using these data , we generated for each colony a family tree where the detailed genealogical relationships and gene-expression histories of corresponding family members are shown ( Figure 2B ) ., Because cells are continuously born throughout the experiment , we aligned them in silico so that their birth times were identical ., In this context , it is natural to define the marginal switch time , τX , a parameter that describes the interval between the birth of a cell X and the moment it eventually becomes fluorescent ( Figure 2C ) ., We normalized each measurement according to its expected likelihood of being observed ( see Figures S4 and S5 , Text S3 ) to account for any biases caused by the cells exponentially dividing throughout our measurement period ., The resulting data fit well to an exponential curve with an effective transition rate of 0 . 12 switches per generation ( Figure 3A , cyan line ) ., The slight discrepancy between data and exponential fit is likely the result of some cells growing out of the focal plane ., The reverse switching distribution , composed of ON cells switching into the OFF state , could not be obtained in this simple way , because in this scenario the long life of the fluorescent proteins makes it difficult to determine the exact moment cells cease production of yellow fluorescent protein ( YFP ) ., This exponentially distributed switching pattern applies to cells chosen at random without regard to genealogy ., However , measuring cells instead on the basis of their family history paints a very different picture ., To demonstrate this difference , we asked how likely a mother and a daughter cell were to have both switched within a small window of time after the cells divided ., We selected all daughter cells with marginal switch times below some value T , and then we measured what percent of their mothers had also switched at or before that time ., The results , summarized in Figure 3B ( open circles ) , show that when a daughter switches shortly after cell division , its mother cell is overwhelmingly likely to do the same ., For example , of the daughters who switch within 400 min of cell division ( about two generations ) , their mothers have approximately a 50% chance of switching in that same period ., This represents a 2-fold increase in the switching rate for a typical unrelated cell ., As T grows to encompass an ever-larger fraction of all daughter cells , the corresponding percent of switching mother cells asymptotically approaches the marginal switch distribution of Figure 3A ( reproduced in black ) , which represents the limit of no genealogical relationship ., As in the marginal switch case above , we are careful to weigh each of these mother-daughter pairs according to how likely we were to experimentally observe them ., To measure the underlying rates governing this process , we examined the possible switching events diagrammed in Figure 3A ., In this simplified view , we assume cell pairs can either switch together into the ON state together at a rate c ( t ) , or independently of one another at a rate r − c ( t ) ., In this way , the total switch rate for any given cell sums to r at all times , as required by the marginal switch distribution ., We assume that the correlations decay with a rate, , which is reminiscent of an Ornstein-Uhlenbeck process ( Figure 3A ) 16 , 28 ., The fixed delay of 20 min is included to account for slow chromophore ( YFP ) maturation as observed in our data ( daughters that switch within the first 20 min after cell division have mothers that always switch ) ., This model includes two free parameters: r , the overall switch rate , and Tc , the characteristic time for the correlation to decay ., A global least-squares fit to both curves ( Figure 3B , red and blue curves ) simultaneously yields ( r = 7 . 0 ± 0 . 5 · 10−4 min−1 = 0 . 12 ± 0 . 01 gen−1 ) and ( Tc = 197 ± 54 min ) ., This decorrelation rate is quite similar to the average cell doubling time of 177 min ( Text S2 and Figure S1 ) , and similar connections between doubling time and decorrelation have been found in other protein regulatory networks 28 ., The above analysis suggests that when cell pairs do switch , they will do so in synchrony ., To demonstrate that this is indeed the case , we turned our focus to the further subset of cell pairs where both cells are observed to switch during the experiment ( and therefore ignoring cases where only one cell in a pair switches ) ., More specifically , we concentrated on three cell relationships: mothers with daughters ( henceforth M-D ) , grandmothers with granddaughters ( GM-GD ) , and older siblings with younger siblings ( S1-S2 ) ., Instead of marginal switching times , which are measured relative to each individual cells time of birth , we chose instead to compute the switch times of both cells relative to the moment when their two respective branches of the family tree first broke apart ., Put another way , this quantifies the amount of time between a switching event and the last moment that these cell lines shared cytoplasm ., The purpose of this approach is to allow us to compare cells that were born at very different times on equal footing , ensuring that switching events are measured relative to the same point for both cells ., For M-D pairs , the time we use is simply the birth of the daughter; for GM-GD pairs , however , it is the birth of the intervening daughter; and for S1-S2 pairs , it is the older siblings birth ., Formally we define the conditional switch time , τX|Y , as the time elapsed between the fluorescing of cell X and the birth of cell Y . When X and Y both refer to the same cell , we recover the marginal switch time ( i . e . , τX|X = τX ) ., Comparing M-D conditional switch times ( Figure 4A ) , we observe nearly synchronous switching that extends at least 300 min and yields a correlation coefficient of ρMD=0 . 87 ( p < 10−45 ) ., GM-GD and S1-S2 pairs ( Figure 4B and 4C ) show somewhat lower correlation coefficients of ρGMGD = 0 . 74 ( p < 10−9 ) and ρSS = 0 . 60 ( p < 10−7 ) , respectively , although the total coefficient for all data combined remains a robust ρTOT = 0 . 8 ( p < 10−62 ) ., The strength and duration of these correlations are surprising and were not found in bacterial 16 , 25 and mammalian 38 studies , except in context of morphological traits 39 ., Like the marginal switch data , these scatter plots should be viewed in the context of finite experimental viewing times , giving weights to points that are inversely proportional to the number of experimental opportunities to have seen them ( Text S3 , Figures S6 and S7 ) ., One dynamic measure for the randomness associated with the distribution is the average square difference of switch times for pairs of cells with comparable mean switch times ( Figure 4D , blue curve ) ., This curve rises rapidly at first , but at longer times it flattens out ., This flattening is likely due , at least in part , to the limited duration of our experiments ( on average 920 min ) , which constrains the scatter distribution to reside in the box shown in Figure 4A–4C ., To understand what this means , it is helpful to compare our results to those obtained using a stochastic Poisson model 40 , where closely related cells are assumed to switch independently of one another and with constant probability in time ( Text S3 , Figures S2 and S8 ) ., To compare directly with our data , we ran the simulation for the same duration as our experiment and included all cell-pair relationships , giving the more complicated curve shown in Figure 4D ( red curve ) ., The ratio of the datas mean square variation to that of the Poisson simulation ( Figure 4E , green curve ) is a measure for how correlated cells remain after a given period of time has passed ., Points below a value of one ( Figure 4E , dashed line ) represent correlated switching behavior , whereas points above it would signify anticorrelated behavior ., For over 600 min , the distribution remains distinctly sub-Poissonian ., Only for the longest measured times are there indications that the cells switch independently of their history , and even this is with large uncertainty ., Put another way , pairs of cells often remain on approximately the same trajectory for several cell divisions , even though cell growth has diluted many of the relevant proteins to a fraction of their original level ., To examine our results at a microscopic level , we constructed a simple model that allows us to probe how the rich correlated switching dynamics arise from a simple regulatory network ., Specifically , we asked whether the stochastic fluctuations of a single regulatory protein in our system could simultaneously explain the observed Poisson switching behavior that is expected for randomly selected individuals and subsequent long–timescale correlations ., One key protein , Gal80p , functions to regulate the expression of all other genes in the network ( Text S1 ) ., When it is present in the nucleus , Gal80p binds in a highly cooperative manner to the transcription factor Gal4p and represses the expression of Gal2p , Gal3p , and YFP ( Volfson et al . , for example , assume a Hill number of 8 between Gal4p and transcription at the GAL1 promoter 33 ) ., Such high levels of cooperativity frequently give rise to steep transfer functions , which can result in switch-like behavior ., This means that even a small decrease in the concentration of Gal80p can cause the transcription rate of downstream genes to increase dramatically from a very small basal rate to a large maximal rate ., Once the downstream protein , Gal3p , begins to be produced , it will lead to sequestering of Gal80p to the cytoplasm , completing the feedback loop and causing the cell to completely switch from the OFF to the ON state ., We constructed a simple model that captures the essential properties of this process ., In our cells , Gal80p is present in very low numbers , and we therefore account for the effects of stochastic production and degradation for this protein ., Protein bursting invariably increases noise levels by amplifying rare events such as changes in promoter activation or mRNA creation and destruction 18 , 22 , 41 ., We assumed that the burst-size distribution was exponential in shape with a mean consistent with the results of Bar-Even et al . , who found an average of 1 , 200 proteins per burst 30 ., We further assumed that the decay rate of the protein is dominated by dilution and therefore set by the division time of the cell ., Finally , we included in our model a nonzero chromophore maturation time of 20 min , as observed in our data ., To account for the cooperativity between Gal80p and Gal4p , we assumed that when Gal80p levels drop below a threshold value a cell rapidly activates gene expression and enters the ON fluorescent state ., In total , the model has only three parameters: ( 1 ) mean number of Gal80p molecules present per cell , ( 2 ) the switching threshold , and ( 3 ) the Gal80p burst size estimated from literature ., We estimated the first two of these parameters by fitting the model to the marginal and conditional switching distributions shown in Figure 3B ., Once the theoretical switching rates were fit to the experimental data , we asked if the model explained the highly correlated switching times observed between related cells ., Without any additional fitting parameters , we predicted the mother ( τM|D ) and daughter ( τD|D ) conditional switching times ( Figure 5F , brown squares ) as well as their mean squared deviation, ( Figure 4E , purple diamonds ) ., These predictions matched remarkably well with the experimental data ( Figure 4E , green boxes; Figure 5F , gray circles ) ., The model therefore predicts that related cells will remain highly correlated in their switching times even though switching events seem to occur in a Poisson manner ., A robustness analysis ( Text S3 , Figure S9 ) suggested a narrow range of possible values with an optimum centered around ( average , threshold ) ∼ ( 2 , 400 proteins , 670 proteins ) ., Bursting events in protein production are often associated with increases of noise in protein levels 18 , 22 ., A counterintuitive aspect of our model is that the correlation observed in cell pairs comes as a consequence of stochastic bursting ., As the burst size is ratcheted up from 12 to the experimentally observed value of 1200 , for example , keeping average protein level and switch rate constant , correlations begin to emerge in the cell-cell scatter plots ( see Figure 5 ) ., The reason for this effect is that the periods between bursting events are dominated by dilution of proteins , a relatively low-noise process ., As the burst size is increased , the time between bursts must increase commensurately , leading to long periods of correlated behavior between cells ., Two cells that start with the same amount of protein will therefore dilute that protein at a similar rate and switch ON ( Figure 5C , black arrows ) at similar times ., Decorrelations can arise when one of the cells experiences a burst of new protein during this decay period ., However , the cell experiencing the burst has a greatly reduced probability of switching ON in a short period of time ., In this event , the cell will generally not be observed to switch at all over the duration of the experiment and consequently does not appear as a significantly decorrelated time-point in the τM|M/τM|D scatter plot ., In recent years , cells within isogenic populations have become increasingly scrutinized as individuals , each with its own original behaviors and gene expression patterns ., What make single cells distinctive , however , are not only the stochastic chemical reactions taking place within them but also their unique family histories ., Here we have shown that a cells decision to dramatically change expression states can hinge directly on this familial background ., We have separated what , on its face , appears to be an exponentially distributed random process into stochastic and genealogically determined subcomponents ., In addition , we show that protein or transcriptional bursting , which are processes that increase total noise in gene expression levels , can unexpectedly create correlated dynamic behavior between related cells , a phenomenon that would be lost in deterministic descriptions ., In the engineered network we used in this study , there is no reason to suppose that the correlations we observe provide an evolutionary advantage to cells ., However , we can speculate that cells might use similar mechanisms to those we describe to coordinate behavior between themselves without relying on complex sensory machinery or physical proximity ., Cells might exploit these architectures to ensure that when a switching event does occur , several other cells will do the same , effectively achieving strength in numbers ., For example , a group of infectious disease–causing cells seeking to confront a host immune system might hypothetically choose to switch together from a slowly growing latent phenotype into an active virulent phenotype in a coordinated but randomly timed attack , thus enhancing their likelihood of sustaining an infection ., Likewise , cells that benefit from cooperative metabolization could similarly benefit from temporally coordinated cooperation ., It will be interesting to see how far similar analysis can be taken in the future and how many other systems might be found to have behavior so strongly influenced by family lineage ., We used the well-characterized GAL network as our model genetic network ( see Text S1 ) ., In wild-type cells , transitions between the ON ( galactose metabolizing ) and OFF ( unable to metabolize galactose ) states is largely determined by the levels of inducers ( e . g . , galactose ) or repressors ( e . g . , glucose ) in the surrounding environment ., To generate a switching phenotype with large dynamic range , we destabilized this in two ways ., First , we removed the negative-feedback loop altogether by replacing the endogenous GAL80 promoter with a weakly expressing , tetracycline-inducible one , PTETO2 ., Second , we grew the cells in the absence of galactose , which fully eliminates the GAL2-mediated positive feedback and weakens the GAL3 feedback ., Even in the absence of galactose , Gal3p has constitutive activity and , in sufficient quantities , can activate the network 42 ., Considering the lower levels of Gal80p in our construct , this constitutive activity is likely a significant factor ., Finally , the state of the network is read with PGAL1-YFP , with fluorescing cells considered ON ., Cells engineered in this way transition between ON and OFF states in a seemingly stochastic fashion ., Cells with this genotype exhibit an extremely broad steady-state expression histogram , with fluorescence values that span more than two orders of magnitude , and the histogram has peaks on both the high and low expression limits , suggesting a bistable system with relatively infrequent transitions between the two states ., Before imaging , cells were grown at low optical density overnight in a 30 °C shaker in synthetic dropout media with 2% raffinose as the sole carbon source ., This neutral sugar is thought to neither actively repress nor induce the GAL genes 43 ., We grew our cells in the absence of tetracycline , so levels of Gal80p were determined by the basal expression level of PTETO2 ., Approximately 12 h later , cells were harvested while still in exponential phase , spun down , and resuspended in synthetic defined ( SD ) media ., Next , cells were transferred to a chamber consisting of a thick agar pad ( composed of the appropriate dropout media and 4% agarose ) sandwiched between a cover glass and slide ., The high agarose density constrains cells to grow largely in a two-dimensional plane ., Fluorescent and phase-contrast images of growing cells were taken at intervals of 20–35 min on 10 different days for over 100 initial progenitor cells ., Image collection was performed at room temperature ( 22 °C ) using a Nikon TE-2000E inverted microscope with an automated state ( Prior Scientific; http://www . prior . com ) and a cooled back-thinned CCD camera ( Micromax , Roper Scientific; http://www . roperscientific . com ) ., Acquisition was performed with Metamorph ( Universal Imaging; http://www . photomet . com ) . | Introduction, Results, Discussion, Materials and Methods | The partitioning and subsequent inheritance of cellular factors like proteins and RNAs is a ubiquitous feature of cell division ., However , direct quantitative measures of how such nongenetic inheritance affects subsequent changes in gene expression have been lacking ., We tracked families of the yeast Saccharomyces cerevisiae as they switch between two semi-stable epigenetic states ., We found that long after two cells have divided , they continued to switch in a synchronized manner , whereas individual cells have exponentially distributed switching times ., By comparing these results to a Poisson process , we show that the time evolution of an epigenetic state depends initially on inherited factors , with stochastic processes requiring several generations to decorrelate closely related cells ., Finally , a simple stochastic model demonstrates that a single fluctuating regulatory protein that is synthesized in large bursts can explain the bulk of our results . | When cells divide , not only DNA but an entire pattern of gene expression can be passed from mother to daughter cell ., Once cell division is complete , random processes cause this pattern to change , with closely related cells growing less similar over time ., We measured inheritance of a dynamic gene-expression state in single yeast cells ., We used an engineered network where individual cells switch between two semi-stable states ( ON and OFF ) , even in a constant environment ., Several generations after cells have physically separated , many pairs of closely related cells switch in near synchrony ., We quantified this effect by measuring how likely a mother cell is to have switched given that the daughter cell has already switched ., This yields a conditional probability distribution that is very different from the exponential one found in the entire population of switching cells ., We measured the extent to which this correlation between switching cells persists by comparing our results with a model Poisson process ., Together , these findings demonstrate the inheritance of a dynamic gene expression state whose post-division changes include both random factors arising from noise as well as correlated factors that originate in two related cells shared history ., Finally , we constructed a model that demonstrates that our major findings can be explained by burst-like fluctuations in the levels of a single regulatory protein . | mathematics, yeast and fungi, biophysics, genetics and genomics | When cells divide, each daughter cell inherits a share of the contents of the mother. If the contents include a regulatory system with a feedback loop, sister cells switch states in synchrony. |
journal.ppat.1005794 | 2,016 | Thermoregulation of Meningococcal fHbp, an Important Virulence Factor and Vaccine Antigen, Is Mediated by Anti-ribosomal Binding Site Sequences in the Open Reading Frame | Neisseria meningitidis is a harmless member of the human nasopharyngeal flora in a significant proportion of healthy individuals 1 ., However in some instances , the bacterium spreads from the upper airway into the systemic circulation , where it can replicate and spread to the rest of the body 2 , especially the cerebrospinal fluid , resulting in meningitis ., Therefore the meningococcus remains an important human pathogen in infants and young adults 3 ., To survive in the human host , the meningococcus has evolved several mechanisms that enable it to evade the immune system 4 ., In particular , the complement system is critical for protection against systemic N . meningitidis infection , evident from the increased susceptibility of individuals with defects in the complement system and findings from a genome wide association study 5 , 6 ., The bacterium evades complement mediated killing by expressing a polysaccharide capsule , sialylation of its lipopolysaccharide , and by binding complement factor H ( CFH ) , the major negative regulator of the alternative complement pathway 7 ., CFH is recruited by high affinity interactions with factor H binding protein ( fHbp ) 8 ., CFH competes with complement factor H related protein-3 ( CFHR3 ) for binding to fHbp on the meningococcus 9; CFHR3 is a competitive inhibitor of CFH binding to the meningococcal surface , and relative levels of CFH and CFHR3 in individuals are likely to determine host genetic susceptibility to meningococcal disease in the general population 9 ., Furthermore fHbp is a surface lipoprotein which is an important component of two vaccines which are now licensed for preventing meningococcal disease 10 , 11 ., Within the upper airway , the bacterium is exposed to gradients in ambient temperature that occur in relation to the anatomical location the phase in the respiratory cycle , and the presence of local inflammation 12 , 13 ., The nasal epithelium contains a complex vascular network with a relatively high blood flow , which forms a heat exchanger , to condition inspired air; the temperature of air entering the nasopharynx rises rapidly from the nostrils , increasing to around 32–34°C by the time it reaches the glottis at the end of inspiration 14 , 15 ., During colonisation , the meningococcus is found both on the epithelial surface as well as deep in the submucosal layer surface 16 , so the bacterium will be will be exposed to a range of temperatures in these sites 17 ., Additionally , during the development of invasive disease , N . meningitidis passes from the lower temperatures in the upper airway to the core body temperature of 37°C or higher with a febrile response to infection 18 ., Therefore , temperature is likely to be an important environmental cue for the meningococcus during colonisation and the development of disease ., We have shown previously that three meningococcal genes , cssA , fHbp , and lst which encode factors contributing to capsule biosynthesis , fHbp , and LPS sialylation , respectively , are subject to thermoregulation ., Of note , the sialic acid-capsule biosynthesis operon is controlled by an RNA thermosensor 19 ., RNA thermosensors are usually located in the 5´-untranslated region ( 5´-UTR ) of an mRNA , and form a secondary structure at lower temperatures that prevents protein translation by blocking access of ribosomes to the nascent mRNA ., As the temperature rises , the secondary structure undergoes a conformational change , exposing the ribosome binding site ( RBS ) , and allowing translation in response to the elevated temperature ., RNA thermosensors have been described in an increasing number of microbes 20 , especially enteric pathogens which are subject to a large fluctuation in temperature upon ingestion by mammalian hosts from the external environment ., Here we describe the mechanisms governing thermoregulation of fHbp ., Expression of fHbp in N . meningitidis is initiated from two transcriptional start sites , resulting in a bi-cistronic transcript ( including the upstream gene , nmb1869 ) , or a mono-cistronic transcript from the fHbp promoter , PfHbp 21 ., It has been shown that transcription from PfHbp is responsive to oxygen limitation , and regulated in an FNR-dependent manner 21 ., We demonstrate that fHbp levels are governed by an RNA thermosensor and demonstrate that sequences in the coding region of fHbp contribute to thermosensing and translation efficiency; two sequences within the fHbp coding region , which are complementarity to the ribosome-binding site , are necessary for the thermoregulation of this key virulence factor and vaccine antigen ., We have shown previously that levels of fHbp in the meningococcus increase following growth at higher temperatures 19 ., To determine whether this affects the level of fHbp on the bacterial surface , we analysed bacteria grown at different temperatures by flow cytometry using αfHbp pAbs ( Fig 1A ) ., The results demonstrate that a rise in temperature is associated with a significant increase in the amount of fHbp on the meningococcus between bacteria grown at 30°C and 42°C ., To establish whether this change is mediated by an alteration in the transcription of fHbp , N . meningitidis was grown to mid-log phase at 30°C , 37°C or 42°C , and fHbp mRNA detected by Northern analysis ., Of note , levels of the bi- and mono-cistronic transcripts of fHbp mRNA were unaffected by temperature , demonstrated by Northern blot analysis ( Fig 1B ) ., However , analysis of samples from the same experiment demonstrate that there was an accompanying clear increase in fHbp levels at higher temperatures ( Fig 1C ) as described previously 19 , indicating that thermal regulation of fHbp occurs at the post-transcriptional level ., To examine whether the increase in fHbp at higher temperatures results from an alteration in protein stability , N . meningitidis was grown at different temperatures to mid-log phase then de novo protein synthesis was inhibited by adding spectinomycin to cultures ., Western blot analyses of samples taken at times afterwards demonstrate that there is no detectable difference in the stability of fHbp in N . meningitidis at 30 and 37°C ( Fig 1D ) ., Taken together , these results indicate that the thermoregulation of fHbp does not result from a change in transcription or protein degradation ., Thermoregulation of fHbp was analysed in Escherichia coli using a plasmid ( pfHbp ) containing the mono-cistronic fHbp and its promoter , PfHbp ( Fig 2A and 19 ) ; results demonstrate that the bi-cistronic mRNA is dispensable for fHbp thermoregulation ., To further define the mechanism of thermoregulation , we examined the level of fHbp mRNA in E . coli harbouring pfHbp grown at 30°C , 37°C and 42°C; consistent with results obtained with N . meningitidis , there was no discernible change in transcript levels following growth at different temperatures ( Fig 2B ) , even though fHbp was subject to clear thermoregulation in the same samples ( Fig 2C ) ., To identify sequences that contribute to thermoregulation , we generated a plasmid ( pfHbpΔPfHbp ) that includes the 5´-UTR of the transcript but lacks its native promoter , PfHbp; in this plasmid , transcription of fHbp occurs from the T7 promoter in the vector ( Fig 2A ) or the SP6 promoter in the opposite orientation ., Again a temperature-dependent increase in fHbp levels was observed in E . coli harbouring fHbp in either orientation in this plasmid ( Fig 2D and S1 Fig ) , demonstrating that PfHbp is not required for fHbp thermoregulation ., In contrast , deletion of the 5´ UTR from pfHbpΔPfHbp ( generating pfHbpΔPfhbpΔUTR ) led to loss of fHbp thermoregulation , indicating that the 5´-UTR of the monocistronic fHbp mRNA is required for thermoregulation in this system lacking the bicistronic transcript ( Fig 2E ) ., To examine the relevance of these findings in N . meningitidis , we substituted the open reading frame ( ORF ) upstream of fHbp ( i . e . nmb1869 , 1065 bp in length ) with the 1120 bp kanamycin resistance marker in N . meningitidis to, i ) leave the promoter for the bi-cistronic transcript intact , and, ii ) disrupt Pfhbp promoter ( Fig 3A and 3B ) , generating strain mutPfHbp-10 ., Northern analysis confirmed loss of the monocistronic fHbp mRNA ( Fig 3C ) , and thermoregulation was still observed in the meningococcus with the bi-cistronic transcript alone ( Fig 3D ) ., Together with the findings in E . coli ( Fig 2 ) , the results provide evidence that the monocistronic PfHbp is dispensable for fHbp thermoregulation ., Next we generated plasmids containing the 5´-UTR with the first one , five or nine codons of fHbp fused to GFP as a reporter ( pfHbp1C-gfp , pfHbp5C-gfp and pfHbp9C-gfp , respectively , Fig 4A ) ; we were unable to generate constructs with more codons , probably because of toxic effects of the fHbp leader sequence in E . coli ., The plasmids lack a T7 promoter 22 , so expression is governed by the native fHbp promoter ., We observed a clear correlation between number of fHbp codons in the plasmid and levels of the GFP reporter , with approximately 3 . 5-fold higher GFP in bacteria harbouring pfHbp9C-gfp compared with fHbp1C-gfp following growth at 37°C ( Fig 4B ) ., The increase in GFP levels observed with pfHbp9C-gfp is not caused by a change in protein or RNA stability ( S2 Fig ) , but instead correlates with the predicted minimum free energy values ( using the Vienna RNAfold package , Fig 4C ) and putative secondary structures ( S3 Fig ) ., To determine whether thermoregulation of fHbp was also observed in a cell free system , in vitro transcription/translation assays were performed with pfHbp1C-gfp , pfHbp5C-gfp and pfHbp9C-gfp ., DNA from these plasmids was transcribed and translated in a continuous manner at 30°C or 37°C , and GFP production assayed by Western blot analysis ( Fig 5A ) ., There was increased GFP generated in assays performed at the higher temperature with pfHbp5C-gfp and pfHbp9C-gfp; however thermoregulation was not observed in assays containing pfHbp1C-gfp ., Overall levels of GFP produced were higher in assays with plasmids harbouring more codons , consistent with sequences in the ORF enhancing translation efficiency ., Therefore , the fHbp9C-GFP construct was used in subsequent experiments ., To further examine the dynamics of fHbp thermoregulation , in vitro transcription/translation reactions were conducted with pfHbp9C-gfp at temperatures between 30°C and 38°C ., Equal amounts of plasmid were used as the substrate for in vitro transcription/translation reactions performed at 30°C , 34°C , 36°C , 37°C and 38°C for one hour , and reaction products were analysed by Western blotting ., The results demonstrate that levels of the GFP reporter rise gradually in response to increasing temperature over a physiological range ( Fig 5B ) similar to the cssA thermosensor 19 ., Next , we performed mutagenesis to identify key nucleotides involved in the formation of putative RNA secondary structures in the fHbp mRNA that mediate thermoregulation ., Several point mutations were introduced upstream of the fHbp GTG start codon in pfHbp9C-gfp ( Fig 6A ) , and their effect examined during growth of E . coli at 30°C , 37°C and 42°C ., Proteins were isolated from the bacteria and levels of GFP expression examined by Western blot analysis ., However , none of the single base substitutions in the 5´-UTR disrupted thermoregulation ( Fig 6B ) , indicating that these changes are insufficient to perturb RNA secondary structures that are necessary for thermosensing ., On further inspection , we identified two copies of a 6 bp repeat ( CUGCCU ) within the fHbp coding region , close to the start codon and potentially able to base-pair with the ribosome binding site ( RBS ) ; we termed these repeats αRBS-1 and αRBS-2 ( Fig 6A ) ., We reasoned that the two cytosines could base-pair with the guanines in the RBS ( AGGAG ) and prevent ribosomes binding to the nascent transcript ., Therefore , site-directed mutagenesis was performed to modify the two cytosines to guanines in each αRBS ( i . e . CUGCCU to CUGGGU ) either singly ( generating mutαRBS-1 and mutαRBS-2 ) , or together ( mutαRBS-1/2 ) ., Constructs were introduced into E . coli , which was then grown at 30°C , 37°C and 42°C , and the level of GFP examined by Western blot analysis ., All three constructs displayed reduced fHbp thermoregulation when compared to the plasmid containing wild-type sequences ., Comparing cultures grown at 30°C or 42°C , the level of GFP increased by 1 . 5-fold and 1 . 6-fold for cells harbouring mutαRBS-1 and mutαRBS-2 , respectively , whereas cells containing mutαRBS-1/2 , showed a 1 . 2-fold increase ( Fig 7A ) ., In contrast , GFP levels increased by 5 . 3-fold with the wild-type sequence in the plasmid ., RecA levels were stable in all conditions , and modification of the αRBS sequences did not affect the stability of the mRNA ( S4 Fig ) ., We also introduced the αRBS mutations into pfHbp , and found that modifications impaired the thermoregulation of fHbp ( S5 Fig ) , although changing the fHbp sequence affected its detection by western blot analysis ., It is likely that the lack of detection of fHbp resulted in changes from the protein sequence ( S5 Fig ) ; mutαRBS-1 results in an Ala to Gly substitution , while mutαRBS-2 alters a Cys to Trp , and a Leu to Val ( S5 Fig ) ., Next we examined whether the α-RBS affect the translation of fHbp ., in vitro translation reactions were performed with equal amounts of total cellular RNA extracted from E . coli strains harbouring the different constructs , and assays conducted at 30°C , 37°C or 42°C ., The in vitro synthesised proteins were extracted and assayed by Western blot analysis ., The wild-type construct showed thermoregulation , whereas the three constructs with altered α-RBS did not display thermoregulation ( Fig 7B ) , similar to results found in E . coli; RecA levels were the same for all conditions examined ., Taken together , these results demonstrate that both α-RBS sequences contribute to fHbp thermosensing ., Additionally , we attempted to introduce the α-RBS mutations into N . meningitidis either under the control of the mono-cistronic or bi-cistronic promoter ( S6 Fig ) ., However similar to results in E . coli ( S5 Fig ) , fHbp was undetectable by Western blot analysis of cell lysates of the meningococcus containing mutαRBS-1/2 , presumably because the amino acid changes affected the processing of the N-terminal region of fHbp and affected its stability ., We considered introducing single C to G mutations into α-RBS1 and/or α-RBS2 , but these nucleotide changes of the α-RBSs are not predicted to impact the overall structure of fHbp mRNA ( S7 Fig ) ., The meningococcus has successfully evolved to survive in the human nasopharynx , which is its only natural habitat ., Colonisation with this bacterium is frequent in young adults ( up to 70% among those living in institutionalised settings such as prisons and barracks ) 1 , and a single strain can persist in the same individual for months in the upper airway 23 ., Within this environment , there are several temperature gradients ., For example , the temperature on the surface of the anterior nares is around 30°C at the end of inspiration , and rises to around 34°C in the posterior nasopharynx and tonsillar region 13 ., Both these sites on the mucosal surface are significantly cooler than the core body temperature of 37°C , where the bacterium replicates during invasive disease ., Additionally , fluctuations in the local temperature will be generated by acute inflammation ( resulting in increased blood flow ) and systemic illnesses , such as influenza , which provoke a febrile response 24 ., We have shown previously that temperature is an important environmental cue for the meningococcus 19 ., The production of fHbp and enzymes necessary for capsule biosynthesis and LPS sialylation are increased as the temperature rises from 30°C through to 42°C ., The enhancement of immune evasion by the bacterium may reflect transition from the cool anterior portion of the nasopharynx ( lined by keratinised squamous cells with little lymphoid tissue ) to the warmer posterior nasopharynx , where the meningococcus is found in tonsillar tissue 16 ., Furthermore , increased temperature could act as a signal of local inflammation 18 or transition to the sub-mucosal layer , where the bacterium will be exposed to immune effectors ., A notable feature of the meningococcus is its capacity to evade exclusion by the human immune system 4 ., Surface structures expressed by N . meningitidis , including Type IV pili , undergo frequent antigenic variation in certain strains 25 , while the serogroup B capsule is a molecular mimic of a human post-translational modification 26 ., Additionally , fHbp mediates high affinity interactions with the complement regulator , CFH using ligand mimicry 8 , and a CFH antagonist CFHR3 9 ., We have shown previously that the biosynthesis of sialic acid-containing capsules is governed by an RNA thermosensor in N . meningitidis 19 ., A fundamental advantage of RNA thermosensors is that they operate at the post-transcriptional level 20 , so do not require a dedicated sensing pathway or de novo transcription to exert their effect ., Therefore , RNA thermosensors are an energetically efficient strategy that offers rapid responses to abrupt changes in the temperature , as seen during the onset of inflammation ., For many bacteria , an increase in temperature is a key signal during their acquisition and ingestion by a mammalian host ., For example , in enteric pathogens , RNA thermosensors are known to modulate expression of a transcriptional regulator , which in turn orchestrates the expression of a suite of genes that enable the pathogen to survive in its new environment 20 ., Examples include the RNA thermosensors in prfA in Listeria monocytogenes 22 , agsA in Salmonella 27 , and lcrF in Yersinia spp ., 28 , 29 ., Thermosensors also govern iron acquisition systems in Shigella and E . coli 30 , and Vibrio cholerae 31 ., We found that the prfA thermosensor operates as an ON:OFF switch 32 , similar to an RNA thermosensor in Cyanobacteria 33 ., The abrupt thermodynamic response of the prfA thermosensor is consistent with Listeria undergoing large temperature fluctuations during its transition from the external environment to the intestinal tract ., However , our in vitro transcription/translation assays reveal that the fHbp thermosensor exhibits a gradual change over a range of physiologically relevant temperatures i . e . 32–35°C 17 , similar to the N . meningitidis capsule thermosensor 19 ., The careful calibration of responses to temperature may be a feature of thermosensors in microbes that exist in close association with thermal gradients within hosts , particularly in the upper airway ., Previous work demonstrated that the expression of fHbp is mediated by bi-cistronic and mono-cistronic transcripts , with the shorter mono-cistronic fHbp transcript controlled by the global regulator of anaerobic metabolism , FNR 21 ., Several lines of evidence demonstrate that the mono-cistronic transcript is sufficient but not necessary for fHbp thermoregulation , which is also detected with the bi-cistronic transcript in N . meningitidis ., Thermoregulation was detected with the mono-cistronic transcript alone in E . coli and using in vitro transcription/translation assays ., Additionally we generated a N . meningitidis mutant lacking the bi-cistronic transcript ( by inserting a kanamycin resistance cassette with a terminator upstream of PfHbp ) ., Even though fHbp levels are lower than in the wild-type strain , consistent with previous work 21 , the strain still exhibited thermoregulation of fHbp ( S6 Fig ) ., The fHbp transcript is not predicted to contain a ROSE element by RNA structure predictions and does not contain a U ( U/C ) GCU sequence close to the RBS , which is often present in this class of thermosensors 20 ., Furthermore , the potential involvement of a temperature regulated sRNA is unlikely as fHbp thermoregulation occurs both in E . coli and in vitro assays ., Instead , our experiments indicate that the mechanism responsible for fHbp thermoregulation depends on two α-RBS sequences located in the fHbp ORF ., Modification of each α-RBS independently reduced thermoregulation , which was virtually abolished when both α-RBSs were altered , indicating some redundancy in their function ., In contrast , mutation of the fHbp 5´-UTR had little or no effect on thermoregulation ., Therefore our results support the model in which the fHbp transcript forms a stem loop at lower temperatures with the RBS occluded by base pairing with one of two α-RBS sequences located in the ORF ., Attempts were also made to introduce the α-RBS mutations into N . meningitidis ., However , we could not detect fHbp in the meningococcus containing modified α-RBSs ( S6 Fig ) , even though fHbp was detected in E . coli using equivalent constructs on a multi-copy plasmid ., The nucleotide changes lead to an alteration in the amino acid sequence of fHbp which might affect lipoprotein localisation and processing , which is distinct in N . meningitidis and E . coli 34 ., We considered other modifications to the αRBS sequences ., However the wobble rule of RNA:RNA binding 35 means that G can bind several bases ( e . g . U and A ) , so their impact on the secondary structure of the RNA would have been uncertain ., The precise contribution of the individual α-RBS sequences to the fHbp RNA thermosensor will be determined in future structural studies ., We also found that the ORF of fHbp contributes to levels of the protein in the meningococcus ., Inclusion of first nine codons of fHbp in GFP fusions resulted in significantly higher protein levels of reporter the compared with constructs containing less sequence ., This was not mediated by changes in protein stability ( S2A Fig ) , and could result from increased efficiency of translation although these findings need to be confirmed in N . meningitidis ., Of note , attempts were made to generate GFP fusions with 12 and 20 fHbp codons in E . coli to define sequences required for maximal fHbp expression ., However , it was not possible to obtain the fHbp12C-gfp and fHbp20C-gfp constructs ., fHbp is a surface lipoprotein bound to the outer membrane via an N-terminal lipid anchor 10 , 11 ., Unlike in the meningococcus , we have been unable to detect fHbp on the surface of E . coli even when expressing the full length protein , demonstrating that there are differences in protein sorting in these two Gram negative bacteria ., It is therefore possible that additional fHbp sequences in gfp fusions cause accumulation of GFP at aberrant cellular sites , impairing bacterial viability ., fHbp thermoregulation may have implications for the effect of vaccines that target fHbp ., Levels of this vaccine antigen in the meningococcus in the upper airway ( at 32–35°C ) may be lower than in assays to assess immune responses within the laboratory , which are typically performed at 37°C 36 , 37 ., Therefore , vaccines that include fHbp might not impose selective pressure on bacteria at the mucosal surface in the upper airway , and offer limited herd immunity 38 ., This is consistent with a recent study demonstrating that immunisation with a vaccine containing fHbp has a limited impact on the acquisition of meningococcal carriage among university students 39 ., Aside from factors involved in immune evasion , RNA thermometers may control other features in N . meningitidis and other bacteria that inhabit the nasopharynx ., Previous studies have focused mainly on the effect of temperature on mRNA levels in the meningococcus 40 ., The identification of further RNA thermosensors will require bio-informatic and proteomic approaches ., Further understanding the mechanisms of thermoregulation could be informative about strategies of immune evasion employed by this important pathogen , its adhesive properties , and the acquisition of relevant nutrients at different sites in the upper airways 41 ., N . meningitidis was grown in Brain Heart Infusion broth ( BHI , Oxoid , 37 g dissolved in 1 L dH2O with 1 g starch ) or on BHI agar ( 1 . 5% w/v ) supplemented with 5% Levinthal’s base ( 500 ml defibrinated horse blood , autoclaved with 1 L BHI broth ) ., Bacteria on solid media were incubated for 16–18 hours at 37˚C with 5% CO2 ., Liquid cultures ( 10 ml ) were inoculated with 109 bacteria and grown at 37˚C with shaking ( 180 r . p . m . ) to an optical density ( O . D . ) measured at 600 nm of ~0 . 5 unless otherwise stated ., E . coli was grown in Luria-Bertani ( LB ) broth ( 2% w/v in dH2O , Oxoid , UK ) or on LB agar ( 1% w/v ) plates ., Liquid cultures of E . coli were grown in 4 ml of media inoculated from a single colony overnight at 37˚C with shaking ( 180 r . p . m . ) ., Overnight grown bacteria were diluted 1 in 100 in media and grown to an Optical Density ( OD ) A600 of ~0 . 5 ., When necessary antibiotics were added to the following final concentrations: carbenicillin , 100 μg ml-1; kanamycin , 50 μg ml-1; rifampicin , 250 μg ml-1 ., The strains and plasmids used in this study are listed in Table, 1 . N . meningitidis was grown in liquid culture to mid ., log phase at 30°C , 37°C or 42°C , prior to fixation for two hours in 3% paraformaldehyde ., Surface localisation of fHbp on N . meningitidis was detected using anti-fHbp V1 . 1 pAbs and goat anti-mouse IgG-Alexa Fluor 647 conjugate ( Molecular Probes , LifeTechnologies ) ., Samples were run on a FACSCalibur ( BD Biosciences ) , and at least 104 events recorded before results were analysed by calculating the geometric mean fluorescence intensity in FlowJo vX software ( Tree Star ) ., A N . meningitidis strain containing only the monocistronic fHbp transcript was generated by insertion of a kanamycin resistance cassette with a Rho-independent terminator upstream of PfHbp ( S6 Fig ) ., Upstream and downstream fragments , and the kanamycin resistance cassette were amplified by from gDNA using the primer pairs fHbp-KM- ( 1 ) -F/fHbp-KM- ( 1 ) -R , fHbp-KM- ( 2 ) -F/fHbp-KM- ( 2 ) -R and fHbp-KM- ( 3 ) -F/fHbp-KM- ( 3 ) -R , respectively ., Details of primers used in this study are given in Table, 2 . PCR products were ligated into pGEM-T ( Promega ) following Gibson Assembly ( NEB ) , then digested with NcoI and NotI ( NEB ) ., Transformation of N . meningitidis strain MC58 was performed as described previously 42 ., N . meningitidis MC58 containing only the bicistronic fHbp transcript was generated by replacing the nmb1869 ( the gene upstream of fHbp ) with the kanamycin resistance gene only , leaving the nmb1869 promoter intact while PfHbp was disrupted by changing the -10 sequence ( TACCATAA to TACCATCC ) ., Upstream and downstream fragments , and the kanamycin resistance cassette together with the modified -10 region mutation were amplified using the primer pairs prom-10-mut1-F/prom-10-mut1-R , prom-10-mut3-F/prom-10-mut3-R , prom-10-mut4-F/prom-10-mut4-R , prom-10-mut5-F/prom-10-mut5-R with genomic DNA as the target , while primers prom-10-mut2-F/prom-10-mut2-R , were used to amplify the kanamycin resistance cassette ., PCR products were ligated into pGEM-T ( Promega ) following Gibson Assembly ( NEB ) , and then digested with NcoI and NotI before being used to transform N . meningitidis ., Site-directed mutagenesis was performed with the Quickchange kit ( Stratagene ) according to the manufacturer’s protocol ., To modify fHbp9C-GFP ( introducing mut1 , mut2 , mut3 , mut4 , mut-αRBS-1 , mut-αRBS-2 and mut-αRBS-1&2 ) , oligonucleotides pairs mut1-F/mut1-R , mut2-F/mut2-R , mut3-F/mut3-R , mut4-F/mut4-R , mut-αRBS-1-F/mut-αRBS-1-R , mut-αRBS-2-F/Mut-αRBS-2-R , mut-αRBS-1&2-F/mut-αRBS-1&2-R were used respectively ( Table 2 ) ., The reaction products were transformed into E . coli DH5α , and constructs were confirmed by sequencing ., Full length fHbp was amplified from N . meningitidis MC58 using primers fHbp ( c ) -F and fHbp ( c ) -R and ligated into pGEMT to yield pfHbp ., Deletions were generated by PCR and the products were then ligated into pGEM-T ( Promega ) and transformed into E . coli DH5α ., The identity of all constructs was confirmed by nucleotide sequencing ., Plasmids containing GFP fusions were generated by amplifying the promoter of fHbp with different lengths of the ORF with fHbp ( TTS ) -U or fHbp-GFP-F with either fHbp-GFP1C-R , fHbp-GFP5C-R or fHbp-GFP9C-R , and ligating the products into pEGFP-N2 ( Clontech ) ., N . meningitidis was grown in liquid culture to mid ., log phase , and RNA was isolated using the RNAeasy Miniprep Kit ( Qiagen , UK ) following the manufacturer’s protocol ., For E . coli , bacteria were grown in liquid media to mid log phase , and RNA isolated using the FastRNA Pro Blue kit ( MP Biomedicals ) according to the manufacturer’s protocol ., The purity and integrity of RNA were determined by gel electrophoresis and spectrophotometry ., For Northern blotting , 20 μg of total RNA was separated on a formaldehyde agarose gel prior to blotting as described previously 43 , then transferred to Hybond N . Membranes were hybridised with 32P-ATP ɣ-labelled DNA fragments Northern blots were developed using a Fuji phosphorImager scanner ., Probes for detecting fHbp and tmRNA were amplified from gDNA with primer pairs fHbp-U/fHbp-D , and tmRNA-U/tmRNA-D , respectively ., To analyse mRNA stability , E . coli containing relevant plasmids was grown in liquid culture to an OD600nm of 0 . 5–0 . 6 then exposed to rifampicin ( 250 μg ml-1 final concentration ) and incubated at 37°C with agitation ., Bacteria were harvested for RNA isolation 0 , 5 , 10 and 20 minutes afterwards ., Blots were probed with a labelled oligonucleotides GGTGCAGATGAACTTCAGGGTCAGCTTGCCGTAGGTGGCATCGCCCTCGC ( to detect GFP mRNA ) , and a tmRNA product amplified from E . coli gDNA with primers Tm1 and Tm2 ( Table 2 ) ., Band intensities of Northern blots were quantified using AIDA image analyzer software , standardised to the tmRNA loading control , and expressed as a ratio to the respective band intensities at t = 0 min ., Cell lysates were prepared by addition of SDS:PAGE loading buffer to an equal volume of bacteria obtained from liquid cultures by centrifugation ., Total protein levels were measured using a Bradford assay and equal amounts of total protein loaded into each lane ., Samples were boiled , then the proteins separated on polyacrylamide gels and transferred to immobolin P polyvinylidine fluoride ( PVDF ) membranes ( Millipore , USA ) using semi-dry transfer ( Biorad , USA ) ., For Western blot analysis , membranes were washed three times in 0 . 05% ( w/v ) dry milk/PBS with 0 . 05% ( v/v ) Tween-20 for 10 minutes , and then incubated with the primary antibody for one hour ., Membranes were washed again three times and incubated for a further hour with a secondary , HRP-conjugated antibody ., Binding was detected with an ECL Western Blotting Detection kit ( Amersham , USA ) and exposed to ECL Hyperfilm ., An α-GFP mouse antibody ( BD-living colors ) was used at a final dilution of 1:8 , 000 ., α-fHbp polyclonal sera was used at a final dilution of 1:5000 ., α-RecA rabbit antibody ( Abcam , UK ) was used at a final dilution of 1:10 , 000 ., As secondary antibodies , goat α-rabbit or α -mouse IgG HRP-conjugated antibody ( Dako , UK ) was used at a final dilution of 1:10 , 000 ., To determine protein stability , translation was prevented by adding spectinomycin ( Sigma , final concentration 100\u2009μg ml-1 ) to bacteria grown to an OD600 = 0 . 5 in liquid media ., Samples were removed at times afterwards for Western blot analysis ., Relative expression was calculated by measuring band intensities with ImageJ software , standardised to signals for loading controls ( i . e . RecA or tmRNA ) , and shown as the ratio , relative to intensity of the control strain or condition ., For the template , DNA was isolated from derivatives of pEGFP-N2 containing one , five or nine codons of fHbp fuse | Introduction, Results, Discussion, Methods | During colonisation of the upper respiratory tract , bacteria are exposed to gradients of temperatures ., Neisseria meningitidis is often present in the nasopharynx of healthy individuals , yet can occasionally cause severe disseminated disease ., The meningococcus can evade the human complement system using a range of strategies that include recruitment of the negative complement regulator , factor H ( CFH ) via factor H binding protein ( fHbp ) ., We have shown previously that fHbp levels are influenced by the ambient temperature , with more fHbp produced at higher temperatures ( i . e . at 37°C compared with 30°C ) ., Here we further characterise the mechanisms underlying thermoregulation of fHbp , which occurs gradually over a physiologically relevant range of temperatures ., We show that fHbp thermoregulation is not dependent on the promoters governing transcription of the bi- or mono-cistronic fHbp mRNA , or on meningococcal specific transcription factors ., Instead , fHbp thermoregulation requires sequences located in the translated region of the mono-cistronic fHbp mRNA ., Site-directed mutagenesis demonstrated that two anti-ribosomal binding sequences within the coding region of the fHbp transcript are involved in fHbp thermoregulation ., Our results shed further light on mechanisms underlying the control of the production of this important virulence factor and vaccine antigen . | The bacterium Neisseria meningitidis is exquisitely adapted to survive in the human host , and possesses several mechanisms to interact with host cells in the upper airway and to circumvent immune responses ., However , the mechanisms that govern the expression of factors that contribute to colonisation and disease are incompletely understood ., In this work , we further characterise how temperature influences the production of factor H binding protein ( fHbp ) by the meningococcus; fHbp recruits human complement proteins to the surface of the bacterium , and is an important vaccine antigen ., We show that thermoregulation of fHbp occurs gradually over a physiological range of temperatures found in the upper airway , the site of colonisation ., This regulation does not require specific meningococcal transcription factors , and sequence analysis indicates that fHbp mRNA forms a secondary structure which could act as an RNA thermosensor ., Additional studies demonstrate that there are two specific sequences within the coding region of fHbp mRNA are important for thermosensing and could base-pair to the ribosome binding site , thus blocking translation of this protein ., As fHbp is thermoregulated , vaccines that target this antigen might not impose a high level of selective pressure on the bacterium at the mucosal surface , thereby limiting herd immunity induce by fHbp containing vaccines . | medicine and health sciences, pathology and laboratory medicine, pathogens, messenger rna, immunology, microbiology, cloning, plasmid construction, vaccines, preventive medicine, physiological parameters, body temperature, dna construction, molecular biology techniques, vaccination and immunization, bacteria, bacterial pathogens, research and analysis methods, public and occupational health, rna structure, neisseria, neisseria meningitidis, medical microbiology, gene expression, microbial pathogens, molecular biology, biochemistry, rna, nucleic acids, protein translation, physiology, genetics, biology and life sciences, organisms, macromolecular structure analysis | null |
journal.ppat.1005261 | 2,015 | Cleavage of a Neuroinvasive Human Respiratory Virus Spike Glycoprotein by Proprotein Convertases Modulates Neurovirulence and Virus Spread within the Central Nervous System | Human coronaviruses ( HCoV ) are enveloped positive-stranded RNA viruses belonging to the family Coronaviridae in the order Nidovirales and are mostly responsible for upper respiratory tract infections 1 ., Being opportunistic pathogens , they have also been associated with other more serious human pathologies , such as pneumonia and bronchiolitis , and even meningitis 2–4 in more vulnerable populations ., Moreover , at least HCoV-229E and HCoV-OC43 are naturally neuroinvasive and neurotropic in humans 5 ., Indeed , we have previously reported that HCoV can infect and persist in human neural cells 6–8 , and in human brains 9 ., Moreover , the OC43 strain ( HCoV-OC43 ) induces encephalitis in susceptible mice , with neurons being the main target of infection 10 , 11 ., Enveloped viruses use different types of proteins to induce fusion of the host-cell membrane to their own in order to initiate infection ., For coronaviruses , the spike ( S ) protein is responsible for cell entry 12 , and was shown to be a major factor of virulence in the central nervous system ( CNS ) for several coronavirus species , including HCoV-OC43 ., We previously reported that persistent HCoV-OC43 infections of human neural cell lines led to the appearance of predominant point mutations in the putative receptor-binding domain of the S glycoprotein gene 13 and that these mutations were sufficient to significantly increase neurovirulence and modify neuropathology in BALB/c mice 14 ., In order to identify amino acid residues in the S glycoprotein that are involved in viral spread within the CNS , we compared the sequence of the gene encoding the viral S protein in the laboratory reference strain HCoV-OC43 ( ATCC VR-759 ) with sequences of the S gene in viruses detected in clinical isolates from sputum of upper and lower respiratory tract of seven children , aged 3 to 36 months , admitted to the University Hospital of Caen , France , in 2003 15 , as well as with all S protein sequences found in the NCBI data bank ., This characterization led to the identification of predominant mutations , including one at the amino acid Gly758 , which introduces a putative furin-like protease cleavage site RRSR↓R758 in the viral S protein 16 ., Several class 1 viral fusion proteins , such as the coronavirus S protein , are proteolytically processed during infection of the host cell , a mechanism that is often essential for the initiation of infection of receptor-bearing cells , tissue tropism and in eventual pathogenesis 17–20 ., Moreover , its cleavage by different types of host proteases , including furin-like proteases designated proprotein convertases ( PCs ) that cleave at paired basic residues 20 are involved in various steps of coronavirus infection 21–23 ., In the present study , we show for the first time , that while the S glycoprotein of the laboratory reference strain HCoV-OC43 ATCC VR-759 is not cleaved by host cell proteases , the sequences of more than 60 clinical isolates reveal a common G758R resulting from a single nucleotide polymorphism ( SNP ) in the S gene ., This creates a functional PC-cleavage site between the S1 and S2 portions of the viral S glycoprotein , thereby modulating viral spread and neurovirulence in susceptible mice , without affecting the neuroinvasive capacities of the virus or its infectivity ( capacity to infect ) of a neuronal cell line ., These results , which suggest that PC-cleavage can be dispensable for efficient infection by HCoV-OC43 , appear surprising compared to other coronaviruses , for which S protein cleavage is required for efficient virus infection 21 , 23 , 24 ., Importantly , our results may help to better characterize the possible adaptation of HCoV-OC43 to the CNS environment , which , in the end , results in a decreased neurovirulence potentially associated to a modified spreading and a more efficient mechanism for the establishment of a persistent infection in human CNS , a phenomenon that could influence the severity of human viral encephalitis or exacerbate neurological degenerative pathologies of unknown etiologies ., We first sought to investigate the potential biological function of the viral G758R mutation located in the HCoV-OC43 S gene between the S1 and S2 domains , detected in the viral S protein of several clinical isolates from human sputum of upper and lower respiratory tract ., Accordingly , we introduced this mutation in the infectious cDNA clone of HCoV-OC43 ( pBAC-OC43FL ) 25 to produce a recombinant mutated rOC/SG758R virus , and we first studied its neuroinvasive and neurovirulent properties compared to reference rOC/ATCC virus ( Fig 1 ) ., For this , 10 day-old BALB/c mice were inoculated by the intranasal ( IN ) route 10 , 14 and survival curves were obtained ( Fig 1A ) ., After infection with the reference virus , over half of BALB/c mice died within the first 15 days post-infection , with symptoms of social isolation and hunched backs ., In comparison , the viral mutant was less neurovirulent , with about 30% of mortality ., Despite this difference in survival , there were no changes in the symptoms induced by the mutant virus compared to reference strain ., Mice were also investigated for variation of weight during infection ( Fig 1B ) as previously described 14: mice infected with mutant virus showed a delay in body weight gain of about 50% at 9 days post-infection compared to control mice ., Comparison of survival curves of mice infected by both variants , coupled with weight variations , suggested that the rOC/SG758R variant was less neurovirulent than reference virus rOC/ATCC after inoculation by the IN route ., To determine whether the slight difference in neurovirulence between the two viruses could be related to differences in viral replication kinetics within the CNS , brains ( Fig 1C ) and spinal cords ( Fig 1D ) were harvested and infectious virus production was evaluated every 2 days over a period of 21 days post-infection ( dpi ) ., Even though the difference in neurovirulence between the two viruses did not correlate with a different amount in production of infectious viral particles in the CNS ( brain and spinal cord ) , there was a delay in viral replication kinetics of the mutant virus compared to reference virus ., Viral spread in mouse brain ( Fig 2 ) was also studied with a focus on the olfactory bulb and the hippocampus regions , because we have previously determined that these regions are primarily infected by the reference virus strain 14 ., At 5 dpi , viral antigens were already present everywhere in the olfactory bulb infected by the reference wild-type virus ( Fig 2A ) , compared to mutant virus for which antigens were only scarcely distributed ., At 7 dpi , the kinetics was restored as the mutant infected this region as efficiently as the reference virus ., In the hippocampus , we observed the same trend: no viral antigens were detected in this region for the mutant virus at 7 dpi , whereas the spread of both viruses was similar at 9 dpi ( Fig 2B ) ., When viruses had spread to all regions of the brain , activation of astrocytes and microglial cells was evident in all infected regions ( S1 Fig ) ., Even though no precise quantitation was performed , a slight increase in the number of astrocytes was observed in the olfactory bulb ( S1A Fig ) and in the hippocampus ( S1B Fig ) of mice infected by the reference virus compared to mutant virus ., Activation of microglial cells was evident in the hippocampus region for both variants at 9 dpi ( S1C Fig ) ., As the mutant virus S protein harbors a SNP present in respiratory clinical isolates , we also evaluated viral dissemination towards the respiratory tract ., Neither infectious virus particles nor viral RNA were detectable in the lungs of all the mice tested ., Having demonstrated that both virus variants retained their neuroinvasive and neurovirulent capacities in our mouse model after intranasal ( IN ) inoculation , we sought to study the spreading and neurovirulent capacities of the two recombinant viruses after intracerebral ( IC ) inoculation , as this route results in a more reproducible infection associated with a better control of viral doses introduced into the brain ., In order to do so , 21 day-old female BALB/c mice were used 11 and experiments were performed by characterizing mouse survival and weight curves , clinical symptoms of encephalitis and viral replication in brain and spinal cord ( Fig 3 ) ., There was a significant difference in survival after inoculation with either virus ( Fig 3A ) : the mutant virus , like the sham control , induced no mortality compared to reference virus , which led to a 20% mortality rate over a period of 21 days ., We then measured the weight of mice during the infection ( Fig 3B ) , and observed that there was a significant delay in body weight gain for the reference virus and the mutant virus compared to the sham control between 7 and 11 dpi , which correlates with the survival curves ., Using a clinical score scale based on neurological symptoms of mice described in the Materials and Methods section 26 , we next studied the clinical symptoms of mice after injection of both variants ( Fig 3C and 3D ) ., The only clinical sign caused by mutant virus was the abnormal flexion of the four limbs ( level 1 ) whereas mice infected by the reference virus developed encephalitis associated with the 4 different levels of clinical scores ., No clinical signs were noted for sham mice ., Taken together , survival and weight curves coupled with the clinical scores indicate that the mutant virus was less neurovirulent than reference virus after IC inoculation in 21-day old mice ., Given our observation that reference virus was more neurovirulent compared to the mutant virus after inoculation by the IC route , we wished to evaluate whether this correlated with a difference in viral replication in the CNS ., Brains and spinal cords were harvested and infectious virus titers were assayed every 2 days for a period of 21 dpi ( Fig 3E and 3F ) ., The difference in neurovirulence did not correlate with a significant difference in the amount of infectious viral particles in the brain ( Fig 3E ) ., However , there was a drastic difference in the production of infectious virus between both variants in the spinal cord ( Fig 3F ) : virus titers of the reference strain ( rOC/ATCC ) were almost identical to what was detected in the brain , whereas the less virulent mutant ( rOC/SG758R ) reached the spinal cord only in one out of thirty infected mice ., In this mouse , an important delay and a production of viral infectious particles close to the limit of detection suggested that mutant virus had difficulty reaching this portion of the CNS ., Histological examination of infected mice revealed that the infected regions were similar following infection by both viruses in the brain , but that the kinetics were different ( Fig 4 ) ., Indeed , as was the case after the IN route of infection , the IC route of infection also led to a delay in viral replication in the olfactory bulb and in the hippocampus , as no viral antigens were detected before 7 dpi for the mutant virus ( compared to 5 dpi for the reference virus ) ., As in 10 day-old BALB/c mice infected IN , when virus had spread to all regions of the brain , activation of astrocytes and microglial cells was evident in all infected regions ( S2 Fig ) ., As seen in 10 day-old mice after IN inoculation , even though no precise quantitation was performed , a slight increase in the number of astrocytes in the olfactory bulb ( S2A Fig ) and in the hippocampus ( S2B Fig ) could be observed in brains of mice infected by reference virus compared to the mutant virus ., The same was observed for microglial cells at 7 dpi in the olfactory bulb ( S2C Fig ) and in the hippocampus ( S2D Fig ) ., In order to further study the role of the G758R mutation on the biology of both HCoV-OC43 variants , we first evaluated the kinetics of viral replication and spread within mixed primary CNS cultures from BALB/c mice over a period of 72 h post-infection ( hpi ) ., Using immunofluorescence , we observed no change in cell tropism , with neurons remaining the main target of infection by both virus variants ( Fig 5 ) , even though astrocytes could also be infected later in the infection ( S3 Fig ) as we previously reported 10 ., Interestingly , we did observe a delay in viral spread in neurons for the mutant virus at 8 and 24 hpi compared to the reference strain ( Fig 5 ) ., Interestingly , even though the infection was shown to be productive for both variants in primary CNS cultures from BALB/c mice , there was a significant increase in the total amount of infectious virus in the cell culture supernatant ( free virus ) between 48 and 72 hpi for the mutant virus compared to the reference virus rOC/ATCC ( Fig 6B ) ., As the G758R mutation creates a putative furin-like cleavage site 16 in the S glycoprotein previously reported to influence viral infectivity 20–22 , 24 , we wished to evaluate whether cleavage was indeed associated with the delayed spreading in neuronal cells , the increased release of infectious virus and eventually with neurovirulence ., As seen in Fig 6 , our data correlated with a much stronger cleavage of the S protein of the rOC/SG758R mutant into S1/S2 fragments , compared to reference virus rOC/ATCC at 24 and 48 hpi ( Fig 6C; whole cell lysate ) , which was even more obvious at 48 hpi in the cell supernatant ( Fig 6D ) ., In order to evaluate whether this cleavage of the viral S protein also took place in human cells , we made use of the differentiated LA-N-5 neuronal cell line described in the Materials and Methods section 27 and showed , first , that the kinetics of viral replication was similar to that observed between both viruses in murine primary cells ( Fig 7A and 7B ) , as there was a significant increase of virus release for the rOC/SG758R mutant and , second , that the cleavage of the S protein into S1/S2 fragments was again predominantly detected in the cell culture supernatant ( Fig 7D ) compared to the protein associated with cells ( Fig 7C ) ., Again , this cleavage was more evident for mutant than for reference virus ., Similar results were obtained with HRT-18 cells ., Even though the S protein of HCoV-OC43 reference virus was present mostly in the uncleaved form , our results also show that there are intermediate size bands between the uncleaved and furin-like cleaved forms of the protein ., These secondary bands may be unspecific degradation products , but we suggest that among these intermediate size fragments seen on SDS-PAGE , there could be a fragment corresponding to the S protein cleaved at a potential alternative site ( S2’ in Figs 6 , 7 , 8 and S4 , the latter showing corresponding overexpositions ) ., In an attempt to determine whether the mutation identified at amino acid 758 ( G758R ) in the viral S protein could indeed create a furin-like cleavage site , we used a cell-permeable general inhibitor of furin-like PCs to investigate the potential involvement of these proteases in the process ., Differentiated LA-N-5 cells were infected in the presence of different concentrations of the decanoylated furin-like inhibitor ( dec-RVKR-cmk ) , and at 48 hpi , proteins in the supernatant were harvested and analyzed ( Fig 8 ) ., As expected , the S glycoprotein of the reference virus was not cleaved at all ( Fig 8A ) whereas the cleavage of the S protein of mutant virus was inhibited in a dose-dependent manner by dec-RVKR-cmk ( Fig 8B ) ., To evaluate whether the kinetics of viral replication was also affected , supernatants were harvested over a period of 48 hpi , and evaluation of infectious viral particles revealed no significant differences ( S5 Fig ) ., As the furin-like inhibitor reduced the cleavage of the S protein for the rOC/SG758R variant , we sought to identify which proprotein convertase ( s ) could play a role in cleavage of the S glycoprotein during infection ., As shown in Fig 8C and 8D , a synthetic peptide containing the sequence of reference virus ( RRSRG ) , was only cleaved by recombinant furin after 15 hours , likely at RRSR↓G ., On the other hand , the synthetic peptide containing the sequence of mutant virus ( RRSRR ) was cleaved in only 30–60 minutes , likely at RRSR↓R , by furin and less so by three additional proprotein convertases: PACE4 , PC5/6 and much less efficiently by PC7 ., Having shown that proprotein convertases are able to cleave the viral S glycoprotein in vitro , we sought to determine whether this cleavage could be associated with a change in viral particle morphology for the rOC/SG758R variant compared to reference rOC/ATCC ., Our observations by transmission electron microscopy ( TEM ) , suggest that the typical coronavirus double crown-shape of the HCoV-OC43 virion was present in two different forms in cell supernatants that were harvested at 48 hpi during infection of mixed primary CNS cultures from BALB/c mice ., Indeed , Fig 9A ( left panel ) represents the first type of morphology , which we named “long” for long S peplomers as measurements of the spike ( S ) and the hemagglutinin-esterase ( HE ) peplomer is shown for the same particle in the right panel ., The same relative length of S and HE proteins were previously determined for other coronaviruses 28 ., The second type of crown morphology , which we named “short” for short S peplomers is presented in Fig 9B ., This short S morphology shows normal HE peplomers of similar length ., For a more accurate characterization of the spike length on viral particles from both variants , we measured the spike length of an equal number of virions ( 10 for each virus ) , for which the crown presented long ( long S ) or short ( short S ) peplomers ( Fig 9C ) ., The average length of the spike associated with a long S type of crown morphology was 24 nm , whereas the average length of the spikes on short S virions was about 15 nm ., This apparent average difference of 9 nm represents a reduction of about 37 . 5% in the total length of the spike , which could presumably play a role in viral infectivity ., Therefore , we next counted the number of viral particles that have a long-S or short-S crown for both viruses ( Fig 9D ) and found a significant difference , which tended to demonstrate that viral particles of the mutant virus were mostly in the short S state ( about 72% ) compared to the reference virus , which showed mainly long S crown ( about 68% ) ., Given that the rOC/SG758R variant was less neurovirulent and presented a delay in dissemination within the mouse CNS compared to rOC/ATCC reference virus , two observations that can relate to a difference in viral infectivity in neuronal cells , we sought to evaluate whether there was a correlation between the morphological differences of the crown of viral particles ( Fig 9A–9D ) and their relative infectivity ( capacity to infect the target cell ) ., No significant differences were found in the ratio of infectious viral particles over total viral particles ( evaluated by the number of viral genome present in viral stocks used for all experiments ) , which establishes itself at 1/200 for both variants ( Fig 9E ) ., Furthermore the amount of viral RNA associated with infected LA-N-5 cells remained the same over a period of 16 hours ( Fig 9F ) , suggesting that attachment and cell entry was similar for both viruses ., Being opportunistic pathogens , HCoV are naturally neuroinvasive and neurotropic in humans 5 ., Herein , making use of our cDNA infectious clone , we show that a single nucleotide polymorphism ( SNP ) naturally found in the S gene of all known HCoV-OC43 contemporary clinical isolates leads to a G758R mutation in the S protein , without significantly affecting the virus neuroinvasive properties and infectivity in cell culture ., However , this mutation was sufficient to modify viral spread and neurovirulence in susceptible mice by modulating the cleavage of the S protein , which appears related to furin-like activity in susceptible neuronal cells ., Even though the rOC/SG758R mutant harbors a SNP present in respiratory clinical isolates , we were not able to detect any viral presence in the respiratory tract of infected mice ., Further studies are underway to try to identify other naturally occurring S mutations that could be important for viral spread to the respiratory tract in mice ., Nevertheless , our results indicate that , despite the difference in neurovirulence , the recombinant virus rOC/SG758R retains its full neuroinvasive properties even though there was a delay in viral spread and in the production of infectious virus ( Figs 1–4 ) ., This phenomenon may in part explain the mutant reduced neurovirulence accompanied by less severe neurological symptoms and a less frequent spread to the spinal cord , as previously reported for other S protein mutants of HCoV-OC43 14 and for the murine coronavirus , MHV 29 ., When viruses had spread to all regions of the brain , the innate immune response was well established , as observed by astrogliosis and microgliosis after both routes of infection where we detected viral antigens 10 , 14 , 30 and S1 and S2 Figs ., The stronger astrogliosis and microgliosis observed after infection by the reference virus may also be related to a faster spread throughout the CNS compared to the mutant virus 14 ., The same difference in viral spread was confirmed in primary cultures of mouse brain cells , where both variants were still infecting neurons as primary targets ( Fig 5 ) , even though astrocytes could also be infected ( S3 Fig ) ., This is in agreement with our previous reports in these cultures 14 and underlines the fact that the change in neurovirulence was not associated with a change in cell tropism as was previously shown for MHV 31 , but could rather be related to a modification in the spread between infected neurons ., The differential neurovirulence and spread could certainly be the consequence of the G758R mutation in the S protein , which introduces a typical furin-like recognition site 16 , 32 probably recognized by cellular proprotein convertases ( PCs ) as this mutation was the only difference found in the whole genome of both recombinant viruses used in the present study ., Proteolytic cleavage of coronaviruses S proteins was characterized several years ago for the murine coronavirus 33 ., Since then , several reports have indicated that PCs appear to be important for cell-cell fusion and/or virus entry into host cells 21–23 , or during transport of the newly assembled virions through the secretory pathway of the producer cell 21 , 34–36 for different coronaviruses including MHV , SARS-CoV and FIPV21 , 23 , 37 , 38 ., The data presented in Figs 6 and 7 clearly show that the S protein harboring the G758R mutation is more easily cleaved during infection ., This S1/S2 cleaved version of the S protein is easily detected in the free virus present in cell culture supernatant but it is barely detectable in cell-associated proteins ., Taken together , these results strongly suggest that this cleavage of S takes place during the late steps of infection , probably during particle assembly and egress , as it was previously shown for MERS-CoV 35 and MHV 39 ., Furthermore , cleavage of the HCoV-OC43 S glycoprotein also has an impact on pathology , as it decreases neurovirulence and spread within the CNS ., It is highly interesting to note that this association between decreased virulence and cleavage of coronavirus S glycoprotein was only suggested for FCoV 32 ., In fact , for other coronaviruses , including the murine ( MHV ) and the bovine coronavirus ( BCoV ) , no clear association was established between S cleavage and virulence 40 , 41 ., The data presented in Fig 8A and 8B strongly suggest that PCs can indeed be involved in the cleavage of HCoV-OC43 S protein during infection of neuronal cells ., Inhibition of furin-like protease ( PCs ) was already demonstrated for other coronaviruses like MERS-CoV and MHV with the same type of inhibitor 21 , 35 ., These results are supported by observation of S protein in reference virus rOC/ATCC ( Fig 8A ) for which the ratio of uncleaved S protein over S1/S2 cleaved form remained equal at all inhibitor concentrations ., In contrast , this ratio increased for mutant virus rOC/SG758R ( Fig 8B ) in a dose-dependent manner ., Results presented in Fig 8C and 8D bring even more interesting new information about which of these proteases could be involved in the actual cleavage ., Indeed , the synthetic peptide harboring the G758R mutation was cleaved with much more efficiency by PCs than the model peptide mimicking the reference virus S protein ., Furthermore , even though furin represented the most efficient convertase , PC5/6 and PACE4 , and to a much lesser extent PC7 , were also able to cleave the synthetic peptide and could therefore cleave the HCoV-OC43 S protein during infection of susceptible cells as it was previously shown for SARS-CoV 42 ., Inhibition of furin-like activity during infection of human neuronal cells , lead us to suggest that PCs ( most notably furin ) are the cognate proteases involved in the S protein cleavage at amino acid 758 ( Fig 8 ) ., The number and morphology of glycoproteins on virions can modulate infectivity for different RNA viruses harboring a class I fusion protein , including other coronaviruses 43–46 ., In the case of HCoV-OC43 , the apparent modification of crown-shaped virions ( Fig 9A–9D ) , associated with the observed differential S protein cleavage , does not seem to increase or decrease viral infectivity ( Fig 9E and 9F ) ., Moreover , inhibition of furin-like-activity did not influence the capacity of the viruses to enter these cells ( S5 Fig ) ., Therefore , even though the S cleavage associated with furin-like activity was shown to influence viral entry for IBV 24 , the HCoV-OC43 S protein cleavage by PCs did not appear to modulate infectivity as it was shown for MERS-CoV 47 ., On the other hand , the modified virion morphology associated with preferential cleavage of the rOC/SG758R HCoV-OC43 variant S protein at the S1/S2 domain interface correlates with a decrease CNS viral spread and neurovirulence in susceptible mice ., Indeed , the delays in spreading in both primary cultures ( Fig 5 ) and within the CNS ( Fig 2 ) were observed despite a more efficient release of infectious rOC/SG758R particles in the cell culture medium as compared to the reference rOC/ATCC virus , a relationship that may seem counterintuitive but is in fact reminiscent of the cell-to-cell mode of propagation prevailing for a growing list of viruses 48 ., For example , HTLV-1 is famously inefficient at spreading through free-virus particles diffusion , the particles remaining instead associated to the plasma membrane from where productive transfer towards target cell occurs 49 , 50 ., By analogy , it is tempting to speculate that S1/S2 HCoV-OC43 spike cleavage limits the amount of particles at the plasma membrane available for a cell-to-cell transfer to naive neurons ., This can explain the difference in kinetics of dissemination between both viruses and the difficulty for mutant rOC/SG758R to reach the spinal cord even though it does disseminate throughout the brain ., This hypothesis may appear in contradiction with the previously documented positive impact of S1/S2 spike cleavage on MHV and SARS-CoV cell-to-cell transfer occurring upon fusion-dependent syncytium formation 21 , 22 , 39 ., Furthermore , even though syncytium formation upon MHV infection has often been linked to S cleavage , in some instances , this type of cell-cell fusion was shown to occur without cleavage of the S protein 51–53 and the MHV-2 strain S protein can be cleaved without being able to induce syncytia 54 ., In fact , regardless of the cleavage status of its S protein , HCoV-OC43 was never able to induce syncytia in any type of cells we studied and we therefore tend to think that distinct , but not mutually exclusive , cell-to-cell propagation mechanisms may prevail among coronaviruses like it does for other enveloped viruses , especially within the CNS , where cell-cell movement of viruses may take place at synapses 48 ., Altogether , these observations suggest that the influence of spike cleavage on coronavirus propagation is not an absolute prerequisite and therefore , cannot per se predict accurately the efficiency of cell-to-cell spread ., The reasons underlying this variable outcome are still unclear but may well reside in the different virus receptors , structural features and attachment factors exploited by coronaviruses ., Given the expected influence of virus spread on neurovirulence , host survival and potentially establishment of CNS viral persistence , further studies are indeed warranted to characterize the underlying mechanisms associated with HCoV-OC43 spread within the CNS ., The SDS-PAGE ( Figs 6 , 7 , 8 and S4 ) shows intermediate size fragments migrating between the uncleaved and furin-like cleaved forms of the S protein that may represent unspecific degradation products ., However , analysis of the S protein gene sequences of HCoV-OC43 , revealed a second putative cleavage site ( S2’ ) between amino acid 899 and 903 ( KASSR ) ., If functional , this second putative cleavage site could be used by other types of cell proteases , including trypsin , TMPRSS and cathepsins 20 , 55–57 to produce a fragment of such molecular weight ., Further studies to characterize the possible involvement of this second putative cleavage site and the identification of host proteases involved in the potential processing of the HCoV-OC43 S protein are ongoing ., Taken together , the results of the current study indicate for the first time that HCoV-OC43 is clearly able to infect neuronal cells and to spread with or without the need for a furin-like S protein cleavage ., The difference in viral spread within CNS and in brain primary cultures , associated with the increase of infectious viral particles in the culture medium for the virus harboring the G758R mutation present in all known clinical isolates , as well as the absence of modification in infectivity between the two viruses , strongly suggests that the PC-activity-associated cleavage of HCoV-OC43 S protein plays a more important role during the egress and viral budding from infected cells , which could influence the mode of viral transmission between CNS cells ., This is of importance to better understand the mechanisms underlying viral spread within the CNS , potentially associated with an adaptation of HCoV-OC43 to this particular environment ., Even though HCoV-OC43 reference strain is highly neurovirulent , we have already shown that its RNA persists in the mouse CNS for up to one year in a significant proportion of infected mice 10 ., Nevertheless , the delayed dissemination and reduced neurovirulence of mutant rOC/SG758R increase host survival and therefore could favor the establishment of CNS viral persistence associated with a potential viral adaptation to the CNS environment , which could result in the selection of better adapted quasi-species , as it was shown for MHV 58 ., In the end , such a persistent infection in the human CNS could , in certain circumstances , be associated with recurrent human encephalitis or neurological degenerative pathologies ., Therefore , the observation that HCoVs are naturally neuroinvasive in both mice and humans 9 , 59 , 60 underlines the need to further characterize viral and cellular determinants of these neuroinvasive properties ., Understanding mechanisms and consequences of virus interactions with the nervous system is essential to better understand potentially pathologically relevant consequences and in the design of diagnostic and therapeutic strategies , including modulation of host proteases such as proprotein convertases ., All animal experiments were approved by the Institutional Animal Care and Use Ethics Committee ( IACUC ) of the Institut National de la Recherche Scientifique ( INRS ) and conform to the Canadian Council on Animal Care ( CCAC ) ., Animal care and used protocols numbers 1304–02 and 1205–03 were | Introduction, Results, Discussion, Materials and Methods | Human coronaviruses ( HCoV ) are respiratory pathogens that may be associated with the development of neurological diseases , in view of their neuroinvasive and neurotropic properties ., The viral spike ( S ) glycoprotein is a major virulence factor for several coronavirus species , including the OC43 strain of HCoV ( HCoV-OC43 ) ., In an attempt to study the role of this protein in virus spread within the central nervous system ( CNS ) and neurovirulence , as well as to identify amino acid residues important for such functions , we compared the sequence of the S gene found in the laboratory reference strain HCoV-OC43 ATCC VR-759 to S sequences of viruses detected in clinical isolates from the human respiratory tract ., We identified one predominant mutation at amino acid 758 ( from RRSR↓ G758 to RRSR↓R758 ) , which introduces a putative furin-like cleavage ( ↓ ) site ., Using a molecular cDNA infectious clone to generate a corresponding recombinant virus , we show for the first time that such point mutation in the HCoV-OC43 S glycoprotein creates a functional cleavage site between the S1 and S2 portions of the S protein ., While the corresponding recombinant virus retained its neuroinvasive properties , this mutation led to decreased neurovirulence while potentially modifying the mode of virus spread , likely leading to a limited dissemination within the CNS ., Taken together , these results are consistent with the adaptation of HCoV-OC43 to the CNS environment , resulting from the selection of quasi-species harboring mutations that lead to amino acid changes in viral genes , like the S gene in HCoV-OC43 , which may contribute to a more efficient establishment of a less pathogenic but persistent CNS infection ., This adaptative mechanism could potentially be associated with human encephalitis or other neurological degenerative pathologies . | Human coronaviruses ( HCoV ) are respiratory pathogens involved in a sizable proportion of common colds ., They have over the years been associated with the development of neurological diseases , given their demonstrated neuroinvasive and neurotropic properties ., The viral spike ( S ) glycoprotein appears to be associated with these neurologic features and is a major factor of virulence for several coronavirus species , including HCoV-OC43 ., To further characterize the role of this protein in neurovirulence and virus spread within the CNS , we sought to identify amino acid residues that may be important for this function ., Our data revealed that one of them , G758R , introduces a functional furin-like cleavage site in the S protein ( RRSR↓R758 ) ., This change in S protein mostly impacts neurovirulence , which seems associated with a modified viral dissemination , without significantly affecting its neuroinvasive capacity ., This mutation , found in all characterized contemporary human clinical respiratory isolates , underlines previous findings that naturally existing field isolates of HCoV-OC43 variants still possess the capacity to invade the CNS where they could eventually adapt and establish a persistent human CNS infection , a mechanism potentially associated with human encephalitis or neurodegenerative pathologies of unknown etiologies . | null | null |
journal.pgen.0030070 | 2,007 | Genetic Basis for Dosage Sensitivity in Arabidopsis thaliana | Most eukaryotic genomes maintain genes in a one-to-one relationship by their syntenic organization on chromosomes ., This normal stoichiometry between chromosomes of a set can sometimes be disrupted , resulting in altered dosage of both genes and their encoded products ., Such disruptions can arise via the nondisjunction of chromatids and chromosomes during mitosis and meiosis and result in uneven chromosome numbers , a condition called aneuploidy ., Trisomy , the most common form of viable aneuploidy is characterized by the presence of one extra chromosome in an otherwise diploid background ., The observation of stereotypical phenotypes for trisomics of each chromosome type illustrated that genetic factors are sensitive to dosage 1–5 ., Indeed , the proper functioning of cells and organisms relies on molecular complexes , which require a delicate balance between components for proper operation 6 ., Even a slight departure from this balance can have dramatic phenotypic or developmental consequences 6 , 7 as exemplified by the many haplo-insufficient genes identified in human as tumor suppressors 8 and as essential or regulatory genes in yeast 7 , 9 and Drosophila 10 , 11 ., In aneuploids , where dosage variations affect whole chromosomes rather than single genes , the consequences can be severe when the copy numbers of many dosage-sensitive genes are altered at once ., Therefore , an alteration of gene dosage such as it occurs in aneuploids typically has unfavorable consequences ., Interestingly , aneuploidy is not always deleterious and can be persistent ., For example , aneuploid cells are normally found in certain tissues such as the brain and the placenta , where they appear to play a functional role 12–15 ., Aneuploidy has been associated with invasive cancer 16 , 17 and controversially proposed to play a causal role in malignancy 18 ., Although cancer is obviously deleterious to the affected organism , somatic selection of aneuploid sectors underscores the fact that dosage imbalance can be advantageous to cells ., Finally , aneuploid individuals are common in plants and in yeast and provide a pool of phenotypic variation not present in the euploid population ., In specific conditions , these phenotypes can be advantageous , and the corresponding aneuploid karyotypes selected ., Such successful aneuploids have been observed both in nature and in industry 19–22 ., Thus , although dramatic alterations of phenotype are associated with aneuploidy , this condition can be compatible with efficient function and even fitness ., There is also tremendous variation for aneuploidy tolerance between different organisms ., For reasons that remain unknown , plants are generally less sensitive than animals to the type of dosage imbalance caused by aneuploidy 17 ., Indeed , in humans , most aneuploidies are embryo-lethal , and the few that are viable are associated with severe developmental defects 23 ., In contrast , trisomics of all chromosome types as well as more complex aneuploid types have been described in several plant species 23–27 ., There is also considerable variation between plant species in the degree of lethality caused by dosage imbalance ., For example , the progeny produced by triploids of different plant species vary in the extent and frequency of aneuploidy ., During triploid meiosis , three sets of chromosomes must be allocated to two poles , producing mostly aneuploid gametes ., The progeny produced by such gametes should consist of a swarm of aneuploid types ranging from near diploid to near tetraploid ., Such a swarm is produced by triploids of certain species , such as A . thaliana 24 , but triploids of other species fail to produce such a range of aneuploids generating instead mostly diploids and near diploids 3 , 23 , 28 , 29 ., Finally , sensitivity to aneuploidy can differ between varieties of the same species ., One such case was reported in tomato in which cherry tomato produced aneuploids with an average of one more extra chromosome than those of a large-fruited tomato variety 30 ., Similar observations were made in barley in which the aneuploids produced by a wild variety carried a higher number of extra chromosomes , were more vigorous , and exhibited higher fertility than the aneuploids produced by a triploid of a cultivated variety 31 , 32 ., In A . thaliana , comparison of identical trisomics of the Columbia ( Col-0 ) and Landsberg erecta ecotypes uncovered differences in fertility and transmission rate of the trisomic chromosome 5 , 33–35 ., A detailed genetic characterization of natural variation for tolerance to aneuploidy and cloning of the responsible loci have so far not been possible ., Recent technological advances in Arabidopsis that combine molecular karyotyping and quantitative genotyping now allow quantitative genetic analysis of aneuploid populations 36 ., Here , we report the first step towards positional cloning of a locus affecting aneuploid survival in Arabidopsis ., We previously investigated the effect of genotype on the rate of aneuploidy production by comparing the karyotype swarms in the progeny of two triploids of A . thaliana ., One genotype , the CCC triploid , was produced from a cross between diploid Col-0 and its synthetically derived tetraploid ( 4x-Col ) ., The other , the CWW triploid , was produced from a cross between diploid Col-0 and the naturally occurring tetraploid Warschau ( Wa-1 ) ., We demonstrated that both of these triploids were fertile and produced a swarm of aneuploid progeny 24 ., Genotype influenced both fertility of the triploids and the composition and performance of their aneuploid swarms ., Additionally , recombinant inbred lines ( RILs ) produced from the progeny of a CWW triploid 37 resolved into two cohorts of near-diploid and near-tetraploid genome contents ( Figure 1 ) ., Genetic analysis of these RILs identified transmission distortion at genetic markers on Chromosome 1 in the near-tetraploid but not the near-diploid lines 24 ., In the present report , we investigated the genetic mechanisms responsible for the ploidy-dependent selection of this locus , which we call SENSITIVE TO DOSAGE IMBALANCE ( SDI ) ., Our results demonstrate that transmission distortion favoring the Wa-1 allele of SDI occurs in genome content classes containing the most severe aneuploids , produced by selfing a triploid ., This indicates a role for SDI in aneuploidy survival , possibly by buffering the dosage-related challenges associated with aneuploidy ., Potential mechanisms consistent with these observations and their evolutionary implications are discussed ., We produced an F2 family from tetraploid CCWW plants ( Figure 2A ) ., Ninety CCWW F2 individuals were genotyped at 11 markers , including nga280 and the linked MN1 . 2 ., No markers exhibited transmission ratio distortion in this population ( Figure 2B ) ., Thus , the polymorphism at SDI is unlikely to be critical for the survival of tetraploids ., As expected from our previous studies of tetraploids 36 , several aneuploid individuals were identified among this CCWW F2 population ( Figure 2A ) ., To investigate whether the SDI allele from Wa-1 was selected in the aneuploid individuals in this F2 population , the percentage of the Wa-1 allele in the aneuploid and tetraploid subpopulations were compared ., Of the 11 markers tested , only MN1 . 2 ( but not nga280 ) exhibited a significant difference between the two subpopulations ., A higher percentage of the Wa-1 allele was present in the aneuploids than in the euploid tetraploid F2 progeny ( t-test p-value = 0 . 0396 ) ( Figure 2C ) ., The previously characterized populations derived by selfing of the CWW triploid 24 were tested for selection at SDI ., In both the CWW F2 population and the near-diploid RILs , the percentage of Wa-1 allele at MN1 . 2 was , on average , higher in the aneuploid individuals than in the euploid individuals ( Figure 3A , inset ) ., This trend was weak and not significant ( t-test p-value = 0 . 38 and 0 . 34 , respectively ) ., A shortcoming of grouping all CWW-produced aneuploids is that differences in the severity of aneuploidy and thus differences in selection for aneuploidy tolerance were not accounted for ., To investigate the relationship between selection of the Wa-1 allele and aneuploidy severity , a quantitative measure of karyotype-dependent selection was developed ., As previously reported , the CWW F2 is a complex swarm of aneuploids of various karyotypes 24 ., This swarm does not match the predicted outcome of triploid meiosis , presumably due to lethality differentially affecting these karyotypes 24 ., The expected frequency of each genome content class was calculated previously ( see Figure 2C in 24 ) ., For each genome content class , the ratio of expected-to-observed frequencies ( Figure 3A ) was used to calculate the aneuploidy selection index ( ASI ) ( see Material and Methods for details ) ( Figure 3B ) ., Negative and positive values for ASI indicated overrepresentation and underrepresentation of a class relative to the expected frequency , respectively ., To test the biological significance of the ASI , we examined the relationship between ASI and seed production in the CWW F2 population ., Both the percentage of plump seed ( as an estimate of seed viability ) and the total number of seed produced by each of the CWW F2 individuals were recorded ( Figure 4A ) ., Their relationship with ASI was determined by regression analyses ( Figure 4B ) ., Both regressions were highly significant ( p-value = 0 . 0002 for seed viability and <0 . 0001 for seed counts ) ., This demonstrated that ASI is a biologically relevant measure correlated with the viability selection acting upon aneuploid classes and predicts the strength of viability selection in the next generation ., Thus , ASI represents an excellent early indicator of karyotype-modulated viability selection ., To test whether the Wa-1 allele at SDI could be linked to increased survival of aneuploid individuals , the relationship between marker genotype and ASI was investigated by regression analysis ., The percentage of Wa-1 allele at three markers , all located at the bottom of Chromosome 1 were significantly associated with ASI ( Table 1 ) ., The most significant association between genotype and ASI was found at MN1 . 2 ( Figure 5 ) ., This association was consistent with a role for SDI in modulating the viability of aneuploids ., Unfortunately , because there is no diploid of Wa-1 , it was not possible to perform a similar analysis on the progeny of a CCW triploid , which could have controlled for potential effects of preferential pairing or segregation ., It is possible that an unidentified meiotic mechanism affected chromosome segregation such that it would be responsible for the effect observed at SDI ., We have observed that the percentage Wa-1 allele at SDI increases with genome content ( regression p-value = 0 . 0006 , r2 = 0 . 12 ) ., Yet , when the effects of genome content and ASI were tested simultaneously , only the effect of ASI ( p-value = 0 . 0096 ) remained significant while that of genome content ( p-value = 0 . 41 ) did not ., This suggested that the apparent increase of percentage Wa-1 allele with genome content was due to the correlation between ASI and genome content ( regression p-value < 0 . 0001 , r2 = 0 . 44 ) and not due to an overall increased percentage Wa-1 allele in disomic gametes ., The percentage Wa-1 allele at MN1 . 2 was lower than expected in the diploid individuals and aneuploids of low genome content ( Figure 5 ) ., This observation was unexpected but appears to fit within a genome-wide phenomenon ., In both the CWW F2 and RIL populations , the percentage Wa-1 allele was on average lower than the expected 66% throughout the genome 37 ., Although we do not have an explanation for this observation , marker MN1 . 2 is not unusual in this respect ., The association between genotype at MN1 . 2 and karyotype was also analyzed using the progeny of pseudo-backcrosses ( pBC ) involving the CWW triploid ., In these populations , one of the two parents is a CWW triploid while the other parent is either diploid Col-0 or its tetraploid derivative ( 4x-Col ) ., ASI values for each chromosome content class were calculated separately for each of the four pBC populations: CWW × Col-0; CWW × 4x-Col; Col-0 × CWW; 4x-Col × CWW ., An association between ASI and marker genotype was again tested by regression ., As observed in the CWW F2 aneuploid swarm , the percentage of Wa-1 allele at MN1 . 2 increased with the ASI but the regression was not significant for any of the four populations studied ( p-values between 0 . 23 and 0 . 92 ) ., Thus , selection for SDI in the progeny of a triploid was only visible in the context of a selfed triploid , where zygotes can be more severely imbalanced and where aneuploidy can have both maternal and paternal origin ., The percentage of Wa-1 allele at SDI increased with our measure of aneuploidy selection in the CWW F2 population ., Thus , selection at SDI in the CWW F2 was karyotype-dependent ., Yet , karyotype-dependent selection at SDI was not significant in the progeny of the pBC ., Comparing the theoretical population of aneuploids produced by a selfed triploid to those produced in the pBC suggests possible explanations for this observation ( Figure 6 ) ., The pBC populations are exclusively composed of moderate aneuploid ( dosage deviation of no more than one chromosome , light blue ) and euploid ( red ) individuals ., These individuals are also present in the triploid F2 distribution , where they only represent a minor proportion ( Figure 6A versus 6C ) ., More extreme aneuploid individuals ( in green in Figure 6 ) , containing two copies of some chromosome types and four copies of others can be formed from a selfed triploid when aneuploid gametes carrying extra copies of the same chromosome types fertilize each other ., In fact , this class of aneuploids constitutes the majority of the possible CWW F2 karyotypes and cannot be produced in pBCs , where one parent contributes only euploid gametes ( Figure 6 ) ., The presence of these extreme aneuploids is supported by the fact that individuals with extreme phenotypes ( e . g . , extreme dwarfism or complete sterility ) were observed in the CWW F2 but not in the pBCs ( unpublished data ) , and that the genome content classes with higher ASI values are those with the highest predicted frequency of severe aneuploids ., The theoretical proportion of these severe aneuploids increases rapidly with increasing chromosome number ., For example , triploids with five chromosome types are expected to produce 56% severe aneuploids ., On the other hand , triploid individuals with ten chromosome types , such as maize , are expected to produce 89% severe aneuploid progeny ., If the severe aneuploid types are more strongly selected against , one would therefore expect that seed yield and viability would be lower in triploids of species with higher chromosome number ., The differences in karyotype distributions between the progeny of a selfed triploid and those from a pBC are not limited to the presence or absence of the extreme aneuploid class ., Also present in the triploid F2 but absent from the pBC , is a second class of moderate aneuploids: those formed following the fusion of two ( instead of one ) aneuploid gametes ( Figure 6A , dark blue ) ., These make up the second largest group of expected progeny in the selfed triploid ., In aneuploid gametes , the production of dosage-sensitive factors needed for the success of fertilization and early development is compromised ., In a fertilization event in which both gametes are aneuploid , the inappropriate production of factors such as those responsible for interploidy seed failure and the endosperm dosage factors 38 are most likely to result in seed failure ., Finally , in a selfed triploid opportunities for selection at the gametophyte generation occur in both pollen and ovules ., This doubles the impact of selection on the haploid generation as compared to the pBCs ., Consistent with the association of SDI selection with aneuploidy severity in the CWW F2 , the highest percentages of Wa-1 allele at SDI are associated with the genome content classes that are predicted to contain the highest percentage of severe aneuploids ( genome content classes 2 . 8 , 3 . 0 , and 3 . 2 ) ., As argued above , selection against severe aneuploids may be an important determinant of genome content distribution in triploid progeny ., Our data suggest that other processes also contribute to it ., If selection against severe aneuploids was the only driving force , one would expect a completely symmetrical bimodal distribution ( encompassing the red , orange , and light blue individuals in Figure 6 ) ., The observed distribution ( Figure 3A ) is bimodal but not symmetrical ., It has been hypothesized that carrying an excess of several types of chromosomes ( such as in genome content classes 2 . 6 or 2 . 8 ) is more deleterious than carrying only one chromosome in excess ( genome content classes 2 . 2 or 3 . 2 ) ., This hypothesis is based on the observation that most regulatory interactions are negative 10 ., Therefore , increasing the number of copies at more loci ( or stated differently , having a minority of loci in relative deficiency ) should increase the probability of negatively affecting loci involved in crucial cellular or developmental processes and that are not located on the additional chromosome copies 10 ., Our results agree with this hypothesis: in each half of the distribution , individuals with lower genome contents are more common than individuals with higher genome content ( Figure 3A ) ., We have shown that the naturally collected tetraploid ecotype Wa-1 produced aneuploid individuals more often than Col-0 36 ., In addition , an allele from Wa-1 is associated with the survival of severe aneuploid individuals ( Figures 2 and 5 ) ., Considering the obvious negative consequences of aneuploidy , is there a counterbalancing advantage that could justify the persistence of such a trait ?, Persistent aneuploidy has been reported in several specific situations where the aneuploid phenotype confers a selective advantage relative to the diploid phenotype ., For example , segmental aneuploidy is frequent in yeast deletion mutants 39 and can confer a growth advantage to the aneuploid cells compared to the euploid ones 39 ., In humans , aneuploid cells have recently been found to be an integral part of the functional pool of neurons and in placental trophoblasts , consistent with a functional role for aneuploidy in these contexts 12–15 ., In plants , it is believed that aneuploidy can play a role in speciation and phenotype evolution 40 ., Additionally , aneuploidy is very frequent in polyploid populations 2 , 40 as well as in the process of polyploid formation through the triploid bridge in species for which triploids are readily produced and are fertile 24 , 41 ., In neopolyploids , aneuploidy may contribute to phenotypic variability 40 and become fixed through selection for advantageous karyotypes ., An allele associated with increased tolerance to dosage imbalance would therefore increase the probability that advantageous karyotypes arise and reproduce successfully ., In addition , it would increase the fertility of triploids produced from unreduced gametes or interploidy hybridization ., This would enhance gene flow between diploid and polyploid subpopulations via a triploid bridge and aneuploid swarms and allow the sharing of alleles arising in either population ., Such recurrent triploid formation between diploid and polyploid populations and repeated formation of polyploid derivatives would therefore hinder polyploid speciation and may explain why multiple karyotypes are often cataloged as a single species based on shared ecological and morphological traits ., The fact that selection acting on SDI is associated with the presence of extreme aneuploids suggests that SDI acts to buffer the effects of dosage imbalance ., This buffering could enhance the survival of aneuploid gametes or that of the fertilization products , either the zygote itself or the endosperm and could be mediated through a number of mechanisms ., Selection at SDI could stem directly from specific dosage effects on one or a few dosage-sensitive genes ., Aneuploidy affects the regulation of genes located both on the varied chromosome and on the rest of the genome 17 , 42–45 ., The intensity of these effects varies from gene to gene and can be positive or negative , illustrating the complexity of the regulatory networks 10 , 17 , 44 , 46 ., Observations in maize and Drosophila have led to the idea of a “dosage-regulatory hierarchy , ” in which the expression of a given gene might be regulated by several dosage-dependent regulators , which in turn are involved in the regulation of several target genes 10 ., This hypothesis would be consistent with the possibility that selection at SDI stems from the misregulation of a specific dosage-sensitive gene linked to SDI ., Karyotype-dependent selection at SDI could also originate from a genome-wide effect mediated by SDI ., Changes in chromosome number affect overall genome maintenance , function , and regulation 47 , 48 ., For example , polyploidy results in variation in epigenetic regulation , as demonstrated by ploidy-sensitive gene silencing and paramutation in Arabidopsis 48–50 ., Epigenetic silencing in polyploids and aneuploids may result directly from dosage imbalance ., This idea is supported by our understanding of the mechanisms underlying dosage compensation in flies , mammals , and worms , which all rely on chromatin remodeling 51 ., In plants , trisomy dependent epigenetic instability has been reported for a transgenic locus in tobacco 52 , 53 ., Similarly , cancerous cells are associated with both aneuploidy and epigenetic modifications 17 , 54 ., In addition , meiotic silencing of unpaired DNA has been demonstrated in Neurospora crassa 55 , in X-chromosome imprinting in C . elegans 56 , and in sex chromosome inactivation in mammals 57 ., It is possible that the presence of unpaired chromosomes during triploid or aneuploid meiosis has similar consequences ., Thus , it is possible that the SDI locus encodes a regulator mediating a genome-wide epigenetic response to dosage imbalance ., The possible involvement of SDI in epigenetic modifications of the genome or in its regulation is an attractive hypothesis that the recent publication of the complete methylome of A . thaliana 58 , 59 , natural variation in A . thaliana methylation level 60 , 61 , and our ability to detect and karyotype aneuploid individuals 36 will help address ., In conclusion , we have established a quantitative measure for aneuploid survival ., Using this trait , we have demonstrated the feasibility of genetic mapping in aneuploids and associated a locus to the variation in aneuploid survival observed in Arabidopsis ., Characterization of the gene ( s ) responsible for SDI should facilitate a better understanding of the mechanisms governing the sensitivity to dosage imbalance and aneuploid syndromes ., All plants were grown on soil ( Sunshine Professional Peat-Lite mix 4 , SunGro Horticulture , http://www . sungro . com ) in a growth room lit by fluorescent lamps ( Model TL80; Philips , http://www . lighting . philips . com ) at 22 ± 3 °C with a 16 h:8 h light:dark photoperiod or in a greenhouse at similar temperatures and light regimes , with supplemental light provided by sodium lamp illumination as required ., Tetraploid lines were described previously 24 ., Col-0 represents the diploid ecotype Columbia , 4x-Col represents tetraploidized Col-0 , and Wa-1 represents the naturally occurring tetraploid ecotype Warschau-1 ( CS6885 ) ., C and W refer to basic genomes or alleles of Col-0 and Wa-1 , respectively ., The CCWW F2 population ( n = 90 ) was obtained by crossing 4x-Col as the seed parent to Wa-1 and allowing three F1 individuals to self pollinate ., The Col-0 × Wa-1 RILs described by Schiff and coworkers 37 were a kind gift from Shauna Somerville ( Carnegie Institution , Stanford University ) ., The CWW triploid plants were generated by crossing Col-0 as the seed parent to Wa-1 ., The CWW F2 population ( n = 109 ) was generated as described 24 ., In order to reduce the complexity of the aneuploid swarm produced by a triploid , pBC populations were generated by crossing CWW triploids to either diploid Col-0 or tetraploid 4x-Col , in both directions 36 ., The four types of pBC populations and the number of individuals analyzed in the context of this report were as follows: Col-0 × CWW ( n = 80 ) , CWW × Col-0 ( n = 102 ) , 4x-Col × CWW ( n = 33 ) , CWW × 4x-Col ( n = 47 ) ., Single siliques were harvested into individual tubes , and all the seeds from each fruit were counted using a dissecting microscope ., To estimate seed viability , seeds were characterized as “plump” if they contained a visible embryo structure at least 20% the size of wild-type seed or “shriveled” if they did not ., On average , five individual siliques were counted for each individual ., Mean values for each group of plants were compared pair-wise using Students t-test and p-values < 0 . 05 were considered significant ., All individuals in the pBC , the CWW F2 , and the RIL populations were analyzed for genome content as previously described 24 , 36 ., Briefly , control A . thaliana samples of known genome content were run before , between , and after experimental samples and used to create a standard curve , allowing us to determine the genome content of our experimental samples ., Individuals were categorized by genome content values expressed as a multiple of the haploid genome content of Col-0 ., On this scale , a 2 . 0 corresponds to a diploid individual , and a 4 . 0 corresponds to a tetraploid individual ., The number of categories was chosen based on how many chromosome number classes were expected ( four classes between diploidy and triploidy as well as four classes between triploidy and tetraploidy ) ., We have previously shown that this method is accurate and precise for A . thaliana aneuploids by comparing our flow results to complete karyotypes ( r2 = 0 . 983 in 36 ) ., Quantitative genotyping was performed as previously described 36 ., Different populations were genotyped at different markers ., The progeny of the pBCs were genotyped at all 12 markers previously described 36 , and the data were used to infer the complete karyotype of each individual 36 ., The CCWW F2 individuals were genotyped at eight of those 12 markers located on four of the five chromosome types , namely MN1 . 5 , MN1 . 6 , MN1 . 7 , MN1 . 2 , nga1126 , nga1145 , MN4 . 2 , and MSAT5 . 19 as well as at the three additional markers nga280 , F5I14 , and nga692 all located on Chromosome 1 ., The CWW F2 individuals were genotyped at MN1 . 5 , MN1 . 7 , MN1 . 2 , nga280 , F5I14 , nga1126 , nga1145 , and MSAT5 . 19 located on three of the five chromosome types ., Finally one marker , MN1 . 2 , located approximately 5 . 7 cM distal to the centromere from nga280 , was added to the Col-0 × Wa-1 RILs genotype data 24 , 37 ., Selection at MN1 . 2 in the near-tetraploid RILs was evaluated according to previous protocols 24 and was identical to that of nga280 ., In the near-diploid population , the percentage of Wa-1 allele at MN1 . 2 was slightly higher than at nga280 ., As a result , comparisons between diploid and tetraploid RILs were not significant after a Bonferroni correction for 11 independent tests but a strong trend was evident ( Fisher Exact test p-value = 0 . 009 ) ., MN1 . 2 spans a deletion polymorphism between Col-0 and Wa-1 , while nga280 amplifies a polymorphic microsatellite repeat ., Because of the technical advantages of using an indel polymorphism for quantitative genotyping 36 , and because of the close linkage to nga280 , we employed MN1 . 2 for the quantitative genetic analyses of the pBCs ., MN markers were designed by identifying short insertions or deletions between Col-0 and Wa-1 present in the sequence database provided by the Magnus Nordborg laboratory ( http://walnut . usc . edu/apache2-default ) 62 ., The sequence and modification of these primers was summarized previously 36 ., Forward primers for markers nga280 , F5I14 , and nga692 24 , 63 were labeled with 6-FAM , NED , and ROX respectively ., For the pBC populations , chromosome doses were inferred from quantitative fluorescent PCR as previously described 36 ., Individuals were categorized depending on their chromosome number ., Individuals with 10 , 15 , or 20 chromosomes were classified as euploids , while all other individuals were classified as aneuploid ., Individuals from the CCWW F2 population were only partially karyotyped as data were only available for four of the five chromosome types ., Individuals for which all quantitative genotypes were consistent with tetraploidy ( n = 72 ) were classified as euploid , while individuals for which the quantitative genotypes for at least one chromosome type indicated the presence of additional or missing chromosomal copies were classified as aneuploid ( n = 18 ) ., For each population , the numbers of Col-0 and Wa-1 alleles were counted ., For example , a CWWW genotype contributed three Wa-1 alleles and one Col-0 allele ., The ratio of C to W within the population was compared to the expected 1:1 ratio using chi-squared tests ., Transmission ratio distortion in the CCWW F2 as compared to the expected 1:1 ratio was tested by chi-square analysis ., Test significance was set at a p-value < 0 . 0083 , equivalent to a Bonferroni correction for a p < 0 . 05 and six independent tests on four chromosome types ., Only the euploid individuals were used for this analysis ( n = 72 ) , to eliminate any possible effect of aneuploidy on genotype ., In order to statistically test the effect of marker genotypes on various traits , genotypes were expressed quantitatively as the percentage of Wa-1 allele ., These values were inferred directly from the genotypes determined using quantitative fluorescent PCR ., For example , the CCCW genotype was assigned a quantitative genotype value of 25 ., These values were used to test the relationship between marker genotype and ASI ( see below ) by regression analysis ., For the CWW F2 populations , regressions associated with p-values < 0 . 01 were considered significant , equivalent to a Bonferroni corrected p < 0 . 05 for five independent tests on three chromosome types ., Similarly , for the pBC populations , regressions associated with p-values < 0 . 005 to control for ten independent tests on ten chromosome types ., A value for ASI was calculated for each genome content class of the CWW F2 population ., The expected frequency of each genome-content class was calculated assuming random assortment of three sets of chromosomes and no selection for or against karyotypes ., For each genome content class , the observed frequency was calculated by dividing the number of individuals observed in that genome content class by the total number of individuals ., The values for ASI for each genome content class were obtained using the following formula: ASI = log 2 ( expected frequency/observed frequency ) ., Using this formula , chromosome number classes that were overrepresented relative to their expected frequencies were assigned negative ASI values , while positive ASI values indicated selection against a chromosome number class ., Some of the genome content classes included few individuals ., We tested an alternative version of ASI in which these classes were excluded from the analysis ., Although the p-values obtained were higher , the percentage of Wa-1 allele was significantly affected by ASI at the same markers as presented in Table 1 ., A similar approach was applied to the pBC populations with the exception that genome content classes were replaced by chromosome number classes since each of the pBC | Introduction, Results, Discussion, Materials and Methods, Supporting Information | Aneuploidy , the relative excess or deficiency of specific chromosome types , results in gene dosage imbalance ., Plants can produce viable and fertile aneuploid individuals , while most animal aneuploids are inviable or developmentally abnormal ., The swarms of aneuploid progeny produced by Arabidopsis triploids constitute an excellent model to investigate the mechanisms governing dosage sensitivity and aneuploid syndromes ., Indeed , genotype alters the frequency of aneuploid types within these swarms ., Recombinant inbred lines that were derived from a triploid hybrid segregated into diploid and tetraploid individuals ., In these recombinant inbred lines , a single locus , which we call SENSITIVE TO DOSAGE IMBALANCE ( SDI ) , exhibited segregation distortion in the tetraploid subpopulation only ., Recent progress in quantitative genotyping now allows molecular karyotyping and genetic analysis of aneuploid populations ., In this study , we investigated the causes of the ploidy-specific distortion at SDI ., Allele frequency was distorted in the aneuploid swarms produced by the triploid hybrid ., We developed a simple quantitative measure for aneuploidy lethality and using this measure demonstrated that distortion was greatest in the aneuploids facing the strongest viability selection ., When triploids were crossed to euploids , the progeny , which lack severe aneuploids , exhibited no distortion at SDI ., Genetic characterization of SDI in the aneuploid swarm identified a mechanism governing aneuploid survival , perhaps by buffering the effects of dosage imbalance ., As such , SDI could increase the likelihood of retaining genomic rearrangements such as segmental duplications ., Additionally , in species where triploids are fertile , aneuploid survival would facilitate gene flow between diploid and tetraploid populations via a triploid bridge and prevent polyploid speciation ., Our results demonstrate that positional cloning of loci affecting traits in populations containing ploidy and chromosome number variants is now feasible using quantitative genotyping approaches . | Each eukaryotic genome is subdivided into a specific number of chromosome types , which in turn are present in a characteristic number of copies , usually the same for all chromosomes ., In the condition called aneuploidy , copy number differs among chromosome types , disrupting their balance and that of their encoded factors ., As a result , aneuploidy is associated with developmental defects and death ., For example , most types of human aneuploids are unviable: the only autosomal aneuploidy compatible with protracted survival , Down syndrome , is caused by the presence of three copies , instead of two , of the very small Chromosome 21 ., In plants , aneuploidy is more common and less deleterious ., This suggests that plants can more easily tolerate the effects of aneuploidy and can be used to investigate them ., Here , we used the model plant Arabidopsis thaliana to produce and investigate populations of aneuploid individuals ., By comparing genetically distinct aneuploid populations , we identified a chromosomal region that is associated with greater aneuploid survival ., Characterizing the genetic mechanism modulating the response to changes in chromosomal dosage and aneuploid survival will help understand how genome organization affects biological processes and why aneuploidy results in such severe developmental defects . | plants, genetics and genomics, arabidopsis, plant biology | null |
journal.pgen.1003774 | 2,013 | An Alteration in ELMOD3, an Arl2 GTPase-Activating Protein, Is Associated with Hearing Impairment in Humans | Many molecular components that are necessary for the development and maintenance of hearing have been discovered by identifying the genes that underlie hearing impairment in humans and mice 1–4 ., Hearing requires the precise and efficient functioning of intricately structured mechanosensory hair cells and supporting cells in the inner ear 3 ., One of the key structures in the mechanotransduction process is the hair cell stereocilium ., Protruding from the apical surface of the hair cells , stereocilia are organized in three rows of decreasing height in a staircase pattern ., Each stereocilium is composed of an actin core that contains cross-linked and bundled γ- and β-actin microfilaments that are uniformly polarized , with the barbed ( positive ) ends localized at the tip ., At the tapered end of the stereocilium , the actin filaments form a rootlet that has been proposed to anchor the structure in the actin-rich meshwork of the cuticular plate 5 ., Interestingly , among the identified hearing loss-associated genes , nineteen encode proteins that interact with actin 6 , 7 ., Numerous studies have demonstrated that actin cytoskeleton-associated proteins are involved in the development , maintenance and stabilization of the stereocilia ( for review , see 6 ) ., Continuous depolymerization of actin filaments at the base and polymerization at the barbed end , termed treadmilling , is thought to be critical to the maintenance of the length of stereocilia 8 , 9 ., However , a recent study demonstrated a rapid turnover of the actin filaments only at the tip of the stereocilia , without a treadmilling process 10 , emphasizing the specific role of proteins at the stereocilia tip in the regulation of actin filaments ., Regardless of the precise site , it is quite clear that the proper regulation of actin dynamics is critical to the generation and maintenance of stereocilia as sensory structures ., The Rho/Rac/Cdc42 family of GTPases is well known as a regulator of actin ., Rho and Rac in the inner ear are involved in the morphogenesis and growth of the otocyst 11 , 12 ., The depletion of Rac1 or both Rac1 and Rac3 in the murine inner ear leads to a shorter cochlear duct with an abnormal sensory epithelium ., Rac may participate in cell adhesion , proliferation , and movements during otic development 11 , 12 ., Several studies have suggested that the activation/inhibition of Rho pathways control the actin depolymerization rate in the outer hair cells 13 , 14 ., Although best known for their roles in the regulation of membrane traffic , there is growing evidence that GTPases in the Arf family can also act via changes in actin ., 15 ., Here , we report the identification of a new deafness gene , which encodes an ELMO/CED 12 domain containing protein , ELMOD3 ., Our biochemical studies demonstrated that ELMOD3 possesses GAP activity against a small GTPase in the Arf family , Arl2 , providing a functional link between Arf family signaling pathways and stereocilia actin-based cytoskeletal architecture ., GAPs are regulators and effectors of the Ras superfamily of GTPases , which are increasingly recognized as providing specificity as well as temporal and spatial regulation to GTPase signaling 16 ., Thus , we believe that the identification of ELMOD3 role in the inner ear provides new insights into signaling processes that are important to hearing in humans ., Family PKDF468 ( Figure 1A ) was recruited after obtaining Institutional Review Board approval and written informed consent ., The family history revealed that the onset of hearing loss was pre-lingual , with no clear vestibular impairment among the deaf individuals ., Pure-tone bone and air-conduction audiometry revealed severe-to-profound mixed ( conductive and sensorineural ) hearing loss in the affected individuals of family PKDF468 ( Figure 1B ) ., Individual V:2 exhibited severe-to-profound mixed hearing loss , with bone conduction thresholds for the right ear displaying a mild downward slope to the severe hearing loss range ., The left ear displayed slightly better bone conduction thresholds with normal values for lower frequencies and a downward slope to severe hearing loss at higher frequencies ( Figure 1B ) ., The audiograms of individual V:5 revealed bilateral severe-to-profound mixed hearing loss , with a large conductive component in both ears ., The bone conduction thresholds exhibited a mild downward slope to moderately severe hearing loss for the right ear and were slightly better on the left , for which the thresholds ranged from borderline normal to moderate hearing loss ranges ( Figure 1B ) ., The clinical evaluation revealed no clear signs of skin , renal , or retinal abnormalities ., To determine the temporal bone malformation , we performed computed tomography ( CT ) scans of two affected ( V:2 and V:11 ) along with a normal hearing sibling ( V:7 ) ., CT scan of individual V:2 revealed all three semicircular and internal auditory canals were intact on both sides ., The middle ear and mastoid appeared well-aerated bilaterally ., Imaging of individual V:11 demonstrated a slightly narrow appearing internal auditory canal on the right side only ., The mastoid air cells and middle ear cleft were well-aerated bilaterally ., The external auditory canal appeared normal as well for both affected individuals ., We initially observed that deafness in family PKDF468 did not co-segregate with short tandem repeat ( STR ) markers for 74 of the reported recessive nonsyndromic deafness loci ( data not shown ) ., We therefore performed a genome-wide linkage analysis and observed that the deafness phenotype of family PKDF468 exhibited significant evidence of linkage to STR markers on chromosome 2p12-p11 . 2 ( Figure 1A ) ., Additional STRs on 2p were genotyped , and haplotype analysis revealed a 0 . 91 Mb linkage interval that was delimited by the markers D2S1387 and D2S2232 ( Figure 1A ) ., Under a recessive model of inheritance , with a disease allele frequency of 0 . 001 and full penetrance , a maximum two-point LOD 17 score of 4 . 74 ( θ\u200a=\u200a0 ) was obtained for the marker D2S2333 ., These results define and delimit DFNB88 Human Genome Nomenclature Committee ( HGNC ) approved locus symbol , a novel recessive deafness locus on chromosome 2p11 . 2 ., The DFNB88 locus partially overlaps with the dominant deafness locus DFNA43 ( Figure 1C ) 18 ., Four known candidate genes were identified within the DFNB88/DFNA43 overlapping linkage region ( Figure 1C ) ., However , Sanger sequencing of these genes did not reveal any pathogenic variants ., Approximately 85% of the disease-causing mutations in Mendelian disorders reside in coding regions or in exon-intron canonical splice junctions 19 ., We therefore performed exome sequencing of an affected individual from family PKDF468 ., The sample was enriched using the NimbleGen SeqCap EZ Exome Library v2 . 0 ( Roche Diagnostics; San Francisco , CA ) , and 100 bp , paired-end sequencing was performed on the Illumina HiSeq 2000 platform ( Illumina ) ., An average of 78 . 94% of bases were sequenced with 20× coverage within the targeted regions ., This yielded a total of 64 , 863 single-nucleotide variants , of which 1 , 928 were not found in the dbSNP133 database ( Table S1 ) ., Based on the recessive mode of inheritance evident in the pedigree , we analyzed genes with homozygous changes and potential compound heterozygous changes ., Additionally , we removed all of the variants that were present in six ethnically matched control samples ( Table S1 ) ., No mutation segregating with hearing loss in family PKDF468 was identified in any of the known deafness-causing genes ( Table S1 ) ., We identified one homozygous transition mutation , c . 794T>C ( p . Leu265Ser ) , in ELMOD3 ( Figure S1 ) on chromosome 2p11 . 3 ( Figure 1C ) that segregated with DFNB88-linked deafness ( Tables S1 and S2 ) ., The c . 794T>C change was not present near the canonical splice junctions and was not predicted to create any aberrant splice site ., However , to confirm that c . 794T>C did not affect splicing of ELMOD3 transcripts , we generated cDNA libraries using the total RNA extracted from the white blood cells of two affected and one normal hearing individual ., Sanger sequencing of sub-cloned PCR products , amplified using primers in either exons 9 and 11 or in exons 9 and 12 ( Figure S2 ) , did not reveal any aberrant splicing product in affected individuals ., Thus , the likely pathogenic affect of the c . 794T>C change is substitution of a highly conserved leucine residue at amino acid position 265 of the human ELMOD3 protein with serine ( Figure 2C ) ., No carrier of c . 794T>C was identified among 524 ethnically matched control chromosomes , in the 1000 Genome database or in the 6500 individuals who are listed in the NHLBI-ESP variant database ( http://evs . gs . washington . edu/EVS/ ) ., Moreover , Polyphen-2 20 , SNPs3D 21 , MutationTaster 22 , PMut 23 , and SIFT 24 predicted that the ELMOD3 mutation would be deleterious ( Table S3 ) ., To further confirm that the p . Leu265Ser allele of ELMOD3 is the only mutation that was associated with hearing loss at the DFNB88 locus , we sequenced the coding , non-coding , and approximately 75 bp flanking sequences of the exon-intron boundaries of all the known candidate genes present within the linkage region in two affected individuals of family PKDF468 ( Figure 1C ) ., No other potentially pathogenic mutation was identified in the affected individuals of family PKDF468 ., Although , ELMOD3 is located outside the reported linkage interval of DFNA43 ( Figure 1C ) 18 , nevertheless we sequenced DNA samples of two affected individuals from the original DFNA43 family and no mutation was found ., We next examined the gene structure and expression of ELMOD3 ., Seven alternatively spliced isoforms of human ELMOD3 were identified ( Figure 2A ) ., Isoform A ( reference sequence NM_ 032213 . 4 ) has a translation initiation codon ( AUG ) in exon 2 , ten coding exons that encode a polypeptide of 391 residues ( Figure S1B ) ., Exons 7 to 11 encode the engulfment and cell motility ( ELMO or CED12 ) domain , which consists of 164 amino acid residues ( Figure S1B; blue box ) ., ELMOD3 isoforms B to D include alternatively spliced exons in the 5′ untranslated region ( UTR ) but harbor the same coding exons and encode identical 381 residue polypeptides that differ from isoform A only at their carboxy termini ( Figures 2A and S1B ) ., The human ELMOD3 isoforms , A and B , share 87% identity , with all the differences clustered near the C-terminus ., Isoforms E , F , and G do not encode the full-length ELMO domain due to alternate splicing of exons in the carboxy terminus ( Figure 2A ) ., The c . 794T>C transition mutation is predicted to result in the substitution of serine for a highly conserved leucine in all of the ELMO domain-containing isoforms of ELMOD3 ( Figures 2A and 2C ) ., In comparison to the human sequence , mouse Elmod3 includes only three known alternatively spliced transcripts ( Figure 2A ) ., RT-PCR and real-time quantitative PCR analysis of multiple human and mouse tissue cDNAs ( Tables S4 and S5 ) revealed the ubiquitous expression of isoforms A/a and B–D/b–c ( Figures S3 and 2B ) ., We also assayed the relative mRNA expression of murine Elmod3 isoforms a and b–c with real-time quantitative RT-PCR of RNA that was extracted from cochlear and vestibular inner ear tissues from postnatal day 0 ( P0 ) , P10 , and P30 C57BL/6J mice ( Figure 2D ) ., The expression of Elmod3 isoform b–c was several-fold higher than isoform a , in both cochlear and vestibular tissues at all of the time points examined ( Figure 2D ) ., Therefore , we focused on the ELMOD3 isoform B for the subsequent biochemical and cellular studies ., The mouse ELMOD3 protein is 70% and 80% identical to the A and B isoforms of human ELMOD3 , respectively , and again the differences are greatest at the C-terminus , although single amino acid changes are scattered throughout the alignments ., To characterize the cellular localization of ELMOD3 , we produced a rabbit polyclonal antiserum against synthetic peptide immunogens from mouse ELMOD3 isoform, b . The sensitivity and specificity of the ELMOD3 antibody was validated in immunoblot and immunofluorescence analyses , in transfected cells and mouse tissues ( Figures S4 and S7 ) ., Our antibodies specifically recognized ELMOD3 isoform b but not murine ELMOD1 , ELMOD2 , or ELMOD3 isoform a ( Figure S4 ) ., We next performed immunolocalization of ELMOD3 in the rat and mouse organ of Corti ( Figures 3 and 4 ) ., In rat cochlea , ELMOD3 immunoreactivity was observed in the stereocilia , kinocilia and cuticular plate of developing hair cells ( Figures 3 and S5 ) ., Before P07 , ELMOD3 staining was very weak in the inner hair cells stereocilia ., By P07 , in auditory hair cells , patchy labeling of ELMOD3 immunostaining was detected along the length of stereocilia ( Figure 3 ) ., In contrast to actin staining , ELMOD3 immunoreactivity was not uniform along the length of each stereocilium and the protein seemed to be excluded from a region near the tip ( Figure 3E ) ., ELMOD3 immunoreactivity was also found in the supporting cells , including pillar and Dieters cells ( Figure 3 ) ., Similar to that seen in the rat ( Figure 3D ) , the stereocilia of inner hair cells in the mouse organ of Corti were more intensely labeled than those of outer hair cells ( Figure 4A ) ., In contrast to the cochlear hair cells , ELMOD3 antibody labeling was observed within the hair cell bodies in the vestibular end organs of both rat and mouse inner ear , but no prominent immunoreactivity was observed in the hair bundles ( Figures 4B–4C and S6 ) ., These observations suggest a unique role for ELMOD3 in cochlear sensory cells and may reflect the functional or structural differences between cochlear and vestibular hair bundles ., No specific immunoreactivity was observed when the primary antibody was omitted ( data not shown ) or when the antibody was pre-incubated with the ELMOD3 peptide antigen ( Figure S7 ) ., We examined LLC-PK1-CL4 epithelial ( CL4 ) cells to understand the mechanism and effect of the hearing loss-associated allele of ELMOD3 ., CL4 cells contain actin-rich microvilli and have been used as in vitro models of stereocilia to examine F-actin and protein dynamics 25 ., We transiently co-transfected GFP-ELMOD3 constructs with Espn constructs , where the latter was used to over-elongate the microvilli at the CL4 cell surface 26 ( Figure 5A–5B ) ., We observed a significant expression of GFP-ELMOD3 in the apical ( microvillar ) plasma membrane twenty-four hours post-transfection ( Figure 5A ) ., We also observed expression of GFP-ELMOD3 in the cytosol of transfected cells ( Figure 5A ) ., In contrast to the wild type protein , the p . Leu265Ser mutation in the ELMO domain yielded a protein that displayed either weak or no labeling in the microvilli of the transfected CL4 cells ., Additionally , the protein appeared to be diffusely located throughout the cytoplasm , with a nuclear concentration in approximately half of the transfected cells ( Figure 5B ) ., Identical results were observed with tdTomato-tagged wild-type and mutant ELMOD3 constructs ( data not shown ) ., To determine the effect of p . Leu265Ser mutation on the localization of ELMOD3 in the mouse inner ear , we performed gene gun-mediated transfection of wild-type and p . Leu265Ser mutant GFP-tagged ELMOD3 cDNA constructs in organotypic cultures of inner ear sensory epithelia of P2 C57BL/6J mice ( Figure 5C–5D ) ., Over-expressed wild-type GFP-ELMOD3 localized along the length of the stereocilia of cochlear hair cells ( Figure 5C ) ., We also observed homogeneous distribution throughout the hair cell bodies ( Figure 5C ) ., Similar to the results that were observed in CL4 cells , GFP-ELMOD3 harboring the p . Leu265Ser mutation failed to target to the stereocilia , and the protein was apparently distributed throughout the cochlear hair cell bodies ( Figure 5D ) ., Taken together , these results support our conclusion that ELMOD3 localizes to actin-based microvilli and stereocilia ( Figure 5 ) but that a point mutation in the ELMO domain can prevent its normal localization and potentially affect its function in the stereocilia ., To further investigate the ELMOD3-actin association , we transfected GFP-tagged ELMOD3 into MDCK cells , which is a highly polarized cell model system ( Figure S8 ) ., Forty-eight hours post-transfection , GFP-ELMOD3 accumulation was apparent at the periphery of the transfected cells near the plasma membrane ( Figure S8A ) ., The expression of GFP-ELMOD3 harboring the p . Leu265Ser mutation in MDCK cells resulted in a protein that failed to target or accumulate at the plasma membrane and instead , appeared to concentrate in the nuclei ( Figure S8B ) ., To determine whether GFP-ELMOD3 associates with the actin cytoskeleton at the plasma membrane ( Figure 6A ) , we treated the cells with cytochalasin D ( cyto-D ) , which is a potent inhibitor of actin polymerization , to disrupt the actin cytoskeleton 27 , 28 ., We hypothesized that if GFP-ELMOD3 associated with the actin cytoskeleton at the cell membrane , then treatment of the cells with cyto-D would also affect ELMOD3 localization ., Indeed , we observed a significant decrease in the GFP-ELMOD3 signal at the cell membrane following disruption of the actin cytoskeleton ( Figure 6B , 6D ) ., Four hours following cyto-D treatment ( i . e . , the recovery period for actin re-polymerization ) 27 , we observed that ELMOD3 re-accumulated at the cell membrane ( Figure 6C , 6D ) ., These results suggest that the localization of ELMOD3 is dependent on the actin cytoskeleton and/or may contribute to a mechanism that supports its maintenance ., To decipher the link between F-actin and ELMOD3 , we performed a two-color stochastic optical reconstruction microscopy ( STORM ) imaging of EGFP-ELMOD3 transfected MDCK cells ., While conventional confocal acquisitions revealed co-localization of ELMOD3 and the actin-cytoskeleton , this high resolution imaging technique allowed us to determine more precisely the relative positions of ELMOD3 and actin ( Figure 7 ) ., ELMOD3 and actin were each found in close apposition to the plasma membrane and in irregularly shaped puncta ( Figure 7C ) ., Many regions of extensive overlap in staining between ELMOD3 and actin , suggest the possibility that a subset of the actin-based structures may contain ELMOD3 but that each protein is also found localized independently of the other at the plasma membrane ( Figure 7C ) ., To test the possibility that ELMOD3 binds to actin-based structures , we performed a high-speed co-sedimentation assay that pellets actin along with its associated proteins ., To obtain a source of purified ELMOD3 , His6-Trigger factor-ELMOD3 ( TF-ELMOD3 ) fusion protein was expressed in bacteria , and the recombinant protein was purified by Ni-NTA chromatography ., The TF-ELMOD3 or control proteins were incubated with polymerized F-actin and subjected to high-speed centrifugation at 150 , 000× g for 1 . 5 hrs ( Figure S9 ) ., TF-ELMOD3 co-sedimented , albeit weakly or incompletely , with F-actin in this assay ( Figure S9 ) ., Under these conditions , the p . Leu265Ser mutation did not significantly impact the level of TF-ELMOD3 that co-sedimented with F-actin ( Figure S9 ) ., Human ELMOD1 and ELMOD2 each possess Arl2 GAP activity 29 ., We therefore investigated whether ELMOD3 also possesses GAP activity against Arl2 ., Previous tests of bacterially expressed human ELMOD3 as either maltose binding proteins or trigger factor fusion proteins were negative , but the homologous preparations of ELMOD1 and ELMOD2 were found to possess very low specific activities as Arl2 GAPs , compared to the preparation purified from bovine tissues ., To obtain a potentially more active preparation of human ELMOD3 , we expressed ELMOD3 and the mutant p . Leu265Ser in HEK293T cells as N-terminal GST-fusion proteins to facilitate protein purification ., Protein expression and purification from ∼108 HEK293T cells that expressed GST-ELMOD3 or the mutant each yielded ∼0 . 6 mg protein ., These preparations were stable at 4°C and against freeze-thaw cycles , as judged by either GAP activity or lack of precipitation ., We expressed and purified GST alone and used it as a negative control in all of our assays ., We also evaluated the effect of cleavage of the GST fusion tag by TEV protease on ELMOD3 activity; no changes in activity in the Arl2 GAP assay were observed compared to un-cleaved proteins ( data not shown ) ., Thus , we believe that the presence of the GST moiety at the N-terminus does not interfere with access to the substrate or with enzymatic activity in our assay ., When we varied the amount of GST-ELMOD3 protein in the Arl2 GAP assay , we observed a dose-dependent response and evidence of saturation at higher protein concentrations ( data not shown ) ., Using the lower concentrations of GST-ELMOD3 to estimate the initial rates of GAP-dependent activity and estimating the purity of the preparation at 50% ( based on visual inspection of Coomassie blue-stained gels ) , we obtained a specific activity of 24 pmol of GTP hydrolyzed/min/mg ( Figure 8 ) ., This specific activity of GST-ELMOD3 as an Arl2 GAP is approximately 32-fold lower than that determined for GST-ELMOD1 and nearly 1000-fold lower than that of GST-ELMOD2 or bovine testes ELMOD2 , which is the most active reported preparation of any Arl2 GAP 29 ., Thus , in contrast to our earlier report that it is inactive , GST-ELMOD3 does exhibit Arl2 GAP activity , and we believe that its lower specific activity when expressed in bacteria likely contributed to the earlier negative findings 29 ., The differences in specific activity among the three human GST-ELMOD preparations from HEK293T cells are predicted to result from differences in substrate specificity , sensitivities to co-activators ( as known for Arf GAPs ) , or both ., Thus , more studies are required to determine whether the biologically relevant substrate of ELMOD3s GAP activity in the inner ear is Arl2 or a related GTPase ., We next assessed the effects of the Leu265Ser point mutation on the Arl2 GAP activity of GST-ELMOD3 ., The mutant protein was expressed at the same levels in HEK293T cells and was purified in the same way , resulting in equivalent amounts of protein , indicating that the protein is equally stable in mammalian cells and in solution ., However , when assayed for Arl2 GAP activity , the mutant was inactive ( Figure 8 ) ., Although we observed small amounts of activity over our no protein control , this level of activity seen for the mutant was not different from that observed with GST alone ( Figure 8 ) ., These activities are so low as to be at or near the lower limits of our assay ., Thus , we can safely conclude that the point mutant has at least a 10-fold lower specific activity than the wild-type protein , but it might be completely inactive as an Arl2 GAP ., Our study revealed that ELMOD3 is important for hearing in humans as a missense mutation in the gene leads to profound hearing impairment ., ELMOD3 belongs to the engulfment and cell motility ( ELMO ) protein family , which includes six known members in mammals ( ELMO1-3 and ELMOD1-3 ) ., Our ex vivo studies reveal that fluorescently tagged ELMOD3 localized with the actin-based microvilli of LLC-PK1-CL4 epithelial cells , in the stereocilia of sensory hair cells of mouse organ of Corti explants , and to a lesser extent to the actin cytoskeleton of MDCK cells , whereas the deafness-associated allele ( p . Leu265Ser ) was deficient in each case ., Similarly , we show that human ELMOD3 possesses Arl2 GAP activity but the mutant has at least a 10-fold loss in activity ., While ELMOD3 antibody reactivity was detected in outer hair cells stereocilia at P02 , more pronounced accumulation of ELMOD3 immunoreactivity was detected in rat cochlear inner hair cell stereocilia only by P12 , which is when hair bundles are in the late phase of maturation ., During this period , the inner hair cell stereocilia undergo a rapid elongation 30 ., The observed staining suggests that ELMOD3 might be necessary for the initial development of the outer hair cell stereocilia or the organization of the bundle in a staircase pattern but may play a different role in the stereocilia of inner hair cells ., Nevertheless , it is tempting to speculate that ELMOD3 may play a role in the maturation or maintenance of the cochlear stereociliary bundle ., Recently , two spontaneous mutations ( rda and rda2J ) in mouse Elmod1 were shown to result in profound deafness and vestibular dysfunction 7 , demonstrating that the function of ELMOD1 is essential for regulating the shape and maintenance of inner ear hair cell stereocilia in mice 7 ., Elmod1 has been shown to be part of a large cluster of genes expressed in the developing inner ear , while Elmod3 level was below the detectable range 31 ., These observations are consistent with findings from other studies , like the SHIELD database , which demonstrated that the level of Elmod3 mRNA is ∼100-fold lower than that of Elmod1 in the developing inner ear ( P0–P1 ) ., Besides stereocilia bundles , immunoreactivity for ELMOD3 was also detected along the kinocilium in developing cochlear hair cells ., Recently , it has been shown that ELMO1 can act at the interface between the actin-cytoskeleton and microtubule network by interacting with ACF7 ( Actin crosslinking family 7 ) 32 ., Moreover , microtubule polymerization depends on Arl2 activity 33 , and we have shown that ELMOD3 exhibits a GAP activity against Arl2 ., Therefore , ELMOD3 expression in the kinocilium might have a role in the assembly of the kinocilium architecture and in pathways regulating planar cell polarity ., The ELMO family proteins are functionally poorly characterized , and more information is currently available for the ELMOs than for the ELMODs , with no structural information available for any of them ., So far , only one activity has been ascribed to ELMOD proteins: we previously reported that recombinant human ELMOD1 and ELMOD2 display in vitro Arl2 GAP activity , whereas ELMOD3 and bacterially expressed ELMO1-3 did not 29 ., More recently , we performed additional phylogenetic and functional analyses of the ELMO domain that led us to re-examine whether ELMOD3 shares the Arl2 GAP activity of ELMOD1 and ELMOD2 ., Our data contrast with the earlier-published claim that ELMOD3 lacks Arl2 GAP activity; we determined that it does indeed possess this activity , albeit at a substantially lower specific activity than that of its two closest human paralogs ., The large differences in specific activities observed in the Arl2 GAP assay may be due to differences in the specificities of ELMODs as GAPs for different GTPases , including the lack of one or more binding partners ( e . g . , one that is perhaps analogous to Dock180 binding to ELMO1 or a co-activator for activity such as has been proposed for COP-I and ArfGAP1 34 ) , and/or the lack of post-translational modification ., Our in vitro experiment revealed that ELMOD3 harboring the p . Leu265Ser mutation , unlike ELMOD3 , has no or few GAP activity against Arl2 ., The lack of Arl2 GAP activity of the mutant may suggest a reduced affinity for the Arl2 GTPase , which may play an important role in ELMOD3 localization ., Elmod3 and Arl2 are expressed in developing mouse cochlear tissues and weakly in vestibular tissues ( Figure S10 ) ., Even though ELMOD3 is active and has been defined as “Arl2 GAPs” , we expect it to be active against other GTPases in the Arf family as well ., We therefore speculate that ELMOD3 functions as a GAP for Arl2 and perhaps other GTPases that participate in actin organization , polymerization or depolymerization in the cochlear hair bundles ., If ELMOD3 is an active GAP for other GTPases , these GTPases are likely to be part of the Arf family given that GAPs are not known to cross family boundaries within the Ras superfamily ., However , it is plausible that ELMOD3 functions in a signaling pathway that includes Arl2 ( or an Arf family GTPase ) , Rac , Rho , and , ultimately , affects the actin cytoskeleton ., Future studies will address the specific roles of ELMOD3 in the development of the inner ear sensory epithelium , cytoskeletal organization , and ELMOD3-mediated signaling pathways ., Revealing the interacting partners , substrate specificities for its GAP activities , as well as the means of regulation of ELMOD3 and other ELMO family proteins , will shed light on the overlapping functions of the Ras superfamily in the inner ear ., These fundamental functions of this unique protein family are likely to be important in all eukaryotic cells ., Family PKDF468 was enrolled in the present study from the Punjab province of Pakistan , and written informed consent was obtained from all participating family members ., The Institutional Review Boards at the Center for Excellence in Molecular Biology ( Pakistan ) , at the National Institute on Deafness and Other Communication Disorders , and at Cincinnati Childrens Hospital ( USA ) approved the present study ., Hearing loss in the affected family members was evaluated using pure-tone audiometry , which tested frequencies that ranged from 125 Hz to 8 kHz ., The family medical history stated that the onset of hearing loss was pre-lingual , and we observed no evidence of vestibular dysfunction or other balance issues using the Romberg and tandem gait tests ., There were no other significant findings from the clinical exam , and the affected members had basic metabolic panel results within the normal range , indicating that they had nonsyndromic hearing loss ., We conducted a genome-wide scan on family PKDF468 using 388 STR markers and performed linkage analysis using GeneMapper software ( Applied Biosystems; Carlsbad , CA ) ., The LOD score was calculated using a recessive model of inheritance assuming a fully penetrant disorder and a disease allele frequency of 0 . 001 ., The primers were designed with Primer3 to sequence all of the coding exons and 75 bp of the exon-intron boundaries of all of the known genes within the DFNB88 locus ( Table S2 ) ., The products were amplified using either Taq polymerase ( Genscript; Piscataway , NJ ) or Amplitaq Gold 360 ( Applied Biosystems ) for the GC-rich regions ., The chromatograms were read using SeqMan software ( DNAStar; Madison , WI ) ., Exome sequencing was conducted on one affected individual from family PKDF468 and was enriched using the Nimblegen SeqCap EZ Exome v2 . 0 Library ( Roche Diagnostics; San Francisco , CA ) ., One hundred base pair paired-end sequencing was performed on an Illumina Hi-Seq 2000 system ., The sequencing data were analyzed following the guidelines that are outlined in the Broad Institutes Genome Analysis Toolkit 35 , 36 ., The row data were mapped using the Burrows Wheeler Aligner 36 , the variants were called using the Unified Genotyper , and the data underwent further processing and quality control 35 , 36 ., Low-quality reads ( less than 10× coverage ) were removed , and the remaining variants were filtered against the dbSNP133 database and all of the known variants in the NHLBI 6500 Exome Variant database that had a minor allele frequency ( MAF ) of greater than 0 . 05% ., We also filtered out additional variants that were observed in six ethnically matched control exomes ., Primers were designed , using Primer3 , to screen the remaining candidate gene variants , and we performed segregation analysis by performing Sanger sequencing of the variants of all of the participating family members ., Human and mouse ELMOD3/Elmod3 isoform-specific primers and TaqMan probes were designed , using Primer3 web-based program , and the transcripts were amplified from human and mouse cDNA libraries ( Clontech Laboratories; Mountain View , CA ) ., Mouse inner ear tissues were harvested from 3 or more C57BL/6J mice at P0 , P10 , and P30 ., The cochlea | Introduction, Results, Discussion, Materials and Methods | Exome sequencing coupled with homozygosity mapping was used to identify a transition mutation ( c . 794T>C; p . Leu265Ser ) in ELMOD3 at the DFNB88 locus that is associated with nonsyndromic deafness in a large Pakistani family , PKDF468 ., The affected individuals of this family exhibited pre-lingual , severe-to-profound degrees of mixed hearing loss ., ELMOD3 belongs to the engulfment and cell motility ( ELMO ) family , which consists of six paralogs in mammals ., Several members of the ELMO family have been shown to regulate a subset of GTPases within the Ras superfamily ., However , ELMOD3 is a largely uncharacterized protein that has no previously known biochemical activities ., We found that in rodents , within the sensory epithelia of the inner ear , ELMOD3 appears most pronounced in the stereocilia of cochlear hair cells ., Fluorescently tagged ELMOD3 co-localized with the actin cytoskeleton in MDCK cells and actin-based microvilli of LLC-PK1-CL4 epithelial cells ., The p . Leu265Ser mutation in the ELMO domain impaired each of these activities ., Super-resolution imaging revealed instances of close association of ELMOD3 with actin at the plasma membrane of MDCK cells ., Furthermore , recombinant human GST-ELMOD3 exhibited GTPase activating protein ( GAP ) activity against the Arl2 GTPase , which was completely abolished by the p . Leu265Ser mutation ., Collectively , our data provide the first insights into the expression and biochemical properties of ELMOD3 and highlight its functional links to sound perception and actin cytoskeleton . | Autosomal recessive nonsyndromic hearing loss is a genetically heterogeneous disorder ., Here , we report a severe-to-profound mixed hearing loss locus , DFNB88 on chromosome 2p12-p11 . 2 ., Exome enrichment followed by massive parallel sequencing revealed a c . 794T>C transition mutation in ELMOD3 that segregated with DFNB88-associated hearing loss in a large Pakistani family ., This transition mutation is predicted to substitute a highly invariant leucine residue with serine ( p . Leu265Ser ) in the engulfment and cell motility ( ELMO ) domain of the protein ., No biological activity has been described previously for the ELMOD3 protein ., We investigated the biochemical properties and ELMOD3 expression to gain mechanistic insights into the function of ELMOD3 in the inner ear ., In rodent inner ears , ELMOD3 immunoreactivity was observed in the cochlear and vestibular hair cells and supporting cells ., However , ELMOD3 appears most pronounced in the stereocilia of cochlear hair cells ., Ex vivo , ELMOD3 is associated with actin-based structures , and this link is impaired by the DFNB88 mutation ., ELMOD3 exhibited GAP activity against Arl2 , a small GTPase , providing a potential functional link between Arf family signaling and stereocilia actin-based cytoskeletal architecture ., Our study provides new insights into the molecules that are necessary for the development and/or function of inner ear sensory cells . | null | null |
journal.pgen.1005994 | 2,016 | A Conserved DNA Repeat Promotes Selection of a Diverse Repertoire of Trypanosoma brucei Surface Antigens from the Genomic Archive | African trypanosomes are protozoan parasites that have dedicated more than 20% of their coding capacity 1 , 2 and 10% total cellular protein content 3 to a single biological function ., To survive in the challenging environmental niche of the mammalian bloodstream , subspecies of Trypanosoma brucei must regularly change their antigenic glycoprotein coat ., In this manner , they are able to escape the antibody-mediated immune response of their host to cause a chronic infection of the bloodstream that results in death of both humans ( African sleeping sickness ) and livestock ( nagana ) if left untreated 4 ., Each parasite’s coat is composed of a densely packed single member of a large family of Variant Surface Glycoproteins ( VSG ) 5 , which are thought to share a conserved membrane-bound structure but are encoded by highly divergent genes 2 ., The T . brucei genome encodes more than 2000 VSG genes and VSG pseudogenes within a genome consisting of 11 megabase chromosomes ( MBC ) , a variable number ( usually 5–10 ) of intermediate chromosomes , and about 100 minichromosomes ( MC ) 2 , 6 ., Yet , only one VSG is expressed at a given time from one of ~15 possible Bloodstream Expression Sites ( BES ) located at the subtelomeres of MBCs 7 ., BESs share a similar sequence and organization , including an RNA polymerase I promoter , a series of Expression Site Associated Genes ( ESAGs ) , a large region of repetitive DNA ( 70-bp repeats ) that precede VSG gene , which is located a short distance upstream of telomere 7 ., While minichromosomal VSGs are also subtelomeric , the majority of the VSG archive is located in VSG arrays on the arms of the MBCs 1 ., Survival of T . brucei in the bloodstream requires the regular activation of silent VSGs from the genomic archive ., Switching from the expression of one VSG coat to the next predominantly occurs by three genetic mechanisms ., A change in the BES being transcribed , resulting in the expression of its subtelomeric VSG , is termed In Situ ( IS ) switching 8 ., Telomeric Exchange ( TE ) is homologous recombination between subtelomeres that results in the exchange of a silent VSG with one in the active BES , retaining both VSG genes 9 ., In contrast , duplicative Gene Conversion ( GC ) , as the name implies , results in the duplication of a silent VSG donor into the active BES and simultaneous deletion of the previously expressed VSG gene 10 ., Unlike IS and TE , which activate silent VSGs already located at subtelomeric sites , GC is the mechanism of VSG switching that permits access to the entire VSG archive ( BES , MC , and MBC arrays ) ., GC is thought to be the predominant mechanism during natural infections 11 and can be activated under laboratory conditions , where rates of switching are low ( ~1x10-5 ) , by increasing subtelomeric DNA breakage at the active BES 12–14 ., Among all switching mechanisms there appears to be a semi-predictable hierarchy of VSG gene selection that begins with the selection of BES-encoded subtelomeric VSGs , followed by non-BES subtelomeric VSGs ( such as those on MCs ) , and finally those from non-telomeric sites in the genome ( loosely organized VSG arrays ) 15 ., Selection of VSGs from other BESs is highly favored during early switch events and is the most common gene selection preference observed under laboratory conditions 12 , 15 ., This is probably because BESs have very similar DNA sequences , including regions of near identity for many kilobases , which would provide ample homology for recombination during gene conversion 7 ., Selection of BES-encoded VSGs alone , of which there are about 15 , would not be expected to support chronic T . brucei infection ., DNA repeat expansions are a common source of genomic translocations ( like gene conversions ) and genomic instability among eukaryotic genomes , and can result in genetic disorders in humans ( reviewed in 16 ) ., Thus , the discovery that the 5’ limit of translocation during VSG switching was a long region of repetitive DNA ( termed the 70-bp repeats based on their approximate length ) led to the predictions that these repeats are possible sites of the DNA lesions that initiate switching , or the source of DNA homology for VSG donor selection in recombination-based switching 17–19 ., Often described as imperfect AT-rich 70-bp repeats , observations that this sequence also occurs proximal to VSGs within the genomic archive bolstered the VSG selection prediction 1 ., Similarly , their predicted role in forming DNA lesions fell into disfavor when it was shown that gene conversion in trypanosomes grown in vitro , albeit at a very low frequency , does not require 70-bp repeats 20 , favoring the proposed role in providing homology for recombination ., Yet , the proposed function of the 70-bp repeats was never experimentally tested ., This was due , in part , to the inability to analyze these events due to low levels of switching that occur under laboratory conditions ., Here , we artificially increase the rate of VSG switching to determine how the 70-bp repeats affect VSG donor selection during gene conversion ., The data presented herein confirm that the 70-bp repeats can function to promote selection of VSGs from throughout the silent repertoire ., In addition , an expanded analysis of the 70-bp repeat sequence enabled us to identify a minimal 70-bp repeat region that promotes archival VSG selection ., In the course of this analysis we also discovered that the 70-bp repeats could have previously unreported affects on the frequency of VSG switching and cell cycle progression ., Furthermore , our data showed that the 70-bp repeats can direct VSG selection away from other BESs , their closest homologs , and toward the genomic archive , which has mechanistic and physiological implications ., Our findings suggest that the 70-bp repeat regions are required for the normal outcomes of VSG switching , and thus the ability of T . brucei to survive in its host during a chronic infection ., To investigate the putative functions of the 70-bp repeats we first subjected the two repeat regions of Lister427 BES1 ( Fig 1A—70 . I & 70 . II ) to fine mapping and the 42 identified repeat sequences were used to produce a consensus sequence logo ( Fig 1B ) ., Similar to previous studies of more limited sample sizes , the repeats were an average of 76-bp ( usually running either 77-bp or 75-bp in length ) and were AT-rich ( 78% ) ., For the sake of consistency within the literature , the 70-bp repeat nomenclature will be maintained 17–19 ., These data support previous work suggesting that the 70-bp sequence is highly conserved 18 and identified two pronounced GC-rich regions ( Region1 and Region 2 ) ., Expanding the analysis to include repeat regions of additional BESs , within both Lister427 7 and TREU927 ( http://www . sanger . ac . uk/resources/downloads/protozoa/trypanosoma-brucei . html ) genomes , showed that this conservation is consistent among T . brucei BES regions ( S1 Fig and S1 Dataset ) ., Thus , in the majority of BESs , a long region of conserved 70-bp sequence is maintained in close proximity to the sub-telomeric VSG gene ., Aside from the BES sequences from these two genomes , direct comparison of the frequency and organization of the 70-bp repeat sequence within available African Trypanosome genomes is limited by the variable quality of each genomic assembly , especially near the subtelomeric regions ( http://tritrypdb . org/tritrypdb/ ) ., Operating within these confines , we sought to determine the prevalence of the 70-bp regions by performing a BLAST analysis of the consensus sequence against each chromosome of the available genomes ( Fig 1C ) ., While the 70-bp repeat sequence was not found in the genomes of South American trypanosome species , which do not undergo antigenic variation , it was abundant within the genomes of T . brucei TREU 927 , T . brucei Lister 427 , T . evansi , and T . brucei gambiense ( a human-infectious subspecies ) ., The abundance of 70-bp repeats in T . evansi ( an emerging pathogen among livestock in the Middle East and Asia ) was anticipated as its genome has extensive similarity with that of T . brucei 21 ., The observation that T . b ., gambiense has fewer 70-bp repeats per chromosome than the other T . brucei subspecies is difficult to interpret as it could be an artifact resulting from the sequencing of its genome ( the genome of another human-infectious form , T . b . rhodesiense , has not been sequenced ) ., In contrast , the absence of the 70-bp repeats from T . congolense and T . vivax could reflect real biological differences in antigenic variation between these very distinct species 22 ., In addition to BESs and megabase chromosomes , VSG-containing contigs from T . brucei Lister 427 minichromosomes contained the 70-bp consensus sequence in the proximity of VSGs ( usually approximately 1 . 5 kb upstream ) ( S1 Table ) 2 ., Thus , the conserved 70-bp repeat sequence identified here is widely distributed among the genomes of African trypanosomes with anticipated positioning in long tracts on BESs and shorter tracts on the megabase and minichromosome arms in the proximity of VSG genes ., The genomic conservation and distribution of this sequence lends support to the hypothesis that the 70-bp repeats contribute to homologous pairing and VSG donor selection during GC 20 ., To test this hypothesis , we sought to genetically manipulate the 70-bp repeats of the active BES and monitor the effects on switching , but were hindered by the naturally low frequency of in vitro switching ( ~1x10-6 ) in the Lister 427 strain ., We therefore established cell lines in which DNA double-stranded breaks ( DSB ) could be induced in the actively expressed BES , to increase the depth of analysis by increasing the frequency of switching by GC 12 , 14 ., An ISceI enzymatic cleavage site was introduced into BES1 proximal to a long region of repeats ( “70 . II-ISceI” , 39 repeat iterations ) , a short region of repeats ( “70 . I-ISceI” , 3 repeats ) and in a repeat deletion mutant ( “Δ70-ISceI” , no repeats ) ( Fig 2A; oligos used for constructs are in S1 Text ) ., The veracity of the ISceI cleavage sites was confirmed by Southern blot analysis and the consistent expression of the ISCEI enzyme among lines confirmed ( Fig 1B and 1C ) ., Five populations of each ISceI-bearing strain ( 70 . II-ISceI , 70 . I-ISceI , or Δ70-ISceI ) were grown for 3 days under normal ( - doxycycline ) or DSB-inducing ( + doxycycline ) conditions , and cells that had switched from their initial VSG ( 427–2 ) to an alternative VSG gene were isolated over magnetic cell-sorting ( MACS ) columns , as described 12 , 13 ( experimental pipeline details S2 Fig ) ., The resulting VSG-switched cells were cloned by limiting dilution and the resulting clones were used to determine both the mechanism of switching ( using established genetic methods 13 , 23 ) and to identify the newly expressed VSG ( using traditional RT-PCR followed by sequence analysis and VSGnome BLAST alignment at http://tryps . rockefeller . edu ) for more than 100 clones from each line ( S2–S4 Tables ) ., As anticipated , based on previous studies 12 , 13 , following DSB induction , all lines switched by GC and preferentially favored the selection of BES-encoded VSG donors ( Fig 2D ) ., Notably , when the 70-bp repeat region proximal to ISceI was long ( 70 . II-ISceI ) , 48% of the selected VSGs arose from minichromosomal ( MC ) or undetermined sites ( UD ) as opposed to homologous BESs ( Fig 2D ) ., In contrast , ISceI break formation proximal to a very small repeat region ( 70 . I-ISceI ) or after repeat deletion ( Δ70-ISceI ) resulted in the selection of BES encoded VSGs in 98% or 100% of clones , respectively ( Fig 2D ) ., Thus , short or deleted 70-bp repeats appeared defective in selecting VSGs from the VSG genomic archive when compared with longer 70-bp repeat regions ., Following a DSB , either naturally occurring or induced , single-stranded DNA is liberated initiating a homology search that is likely resolved by break-induced replication ., Genetic analysis of individual switched clones can determine the extent of DNA transferred from the donor site into the active BES during GC ., One of the most common switching events observed was between BES1 and BES7 , resulting in the expression of VSG427-3 ., Using a BES7 probe upstream of the VSG , clones that have recombined VSG427-3 into BES1 will form a new band ( upon appropriate restriction digestion ) whose length indicates the region of BES7 transferred during GC ., The resulting data indicate how GC affects 70-bp repeat maintenance or the recovery of defective repeat regions ( Fig 3 ) ., Clones arising from a BES1 with normal 70-bp repeats ( Fig 3A—70 . II-ISceI ) showed a variety of outcomes that included the addition of no new repeats ( 5_H9 = 6 . 6 kb ) , partial addition of BES7 repeats ( 2_E5 ~8 kb & 5_F6 > 10 kb ) , or the translocation of full length BES7 repeats ( 2_B8 > 12 kb ) ., In contrast , when BES1 harbors no 70-bp repeats ( Fig 3B—Δ70-ISceI ) the full region of BES7 repeats was consistently incorporated into BES1 during switching ( 1_A4 , 3_A2 , & 3_B10 > 12 kb ) ., In one clone it appears that a region larger than the BES7 repeats was incorporated into BES1 ( 3_E9 ) ; similar long-range recombination events have been reported during GC switching in other studies 23 ., It should be noted that the determination of the precise lengths of the regions transferred from BES7 to BES2 is hindered by the fact that the exact length of the repeats encoded in BES7 is unknown ., Thus , we observe that the 70-bp repeat region in the active BES can be repopulated , maintained , or extended during GC-based recombination with another BES ., The growth rate and frequency of VSG switching following DSB induction could affect the number of VSG donors selected ., To verify that the VSG donor selection phenotypes reported in Fig 2 were dependent solely on the effect of the 70-bp repeat regions , cellular growth and VSG switching were monitored in these lines ., Following doxycycline induction , all lines harboring an ISceI site in BES1 displayed a growth defect when compared to the parental line ., For strains with intact 70-bp repeats ( Fig 4A—70 . II-ISceI blue lines ) the delay in growth was modest , yet DSB formation in the 70-bp deletion mutant ( Δ70-ISceI ) exacerbated a pronounced preexisting growth defect ( Fig 4A—Red lines ) ., We predicted that the growth defect observed without doxycycline induction resulted from leaky expression of the ISCEI enzyme ( a known complication of expression from the rDNA spacer 24 ) , and tested this prediction using a 70-bp repeat deletion mutant that did not harbor the ISCEI enzyme ( Δ70-NO ISCEI ) ., Deletion of the repeats from BES1 did not result in a growth defect in the absence of the ISCEI ( as anticipated from previous work on a similar construction 20 , 25 ) ., Thus , the observed growth defects in ISCEI-expressing lines appear to result from DSB formation in the active BES , which was most pronounced when the 70-bp repeats were deleted ., Because DSB formation can activate a cell cycle checkpoint ( reviewed in 26 , 27 ) and the Δ70-ISceI cell line has a growth defect , the effects of DSB formation on the cell cycle were examined in these lines ., Cells harboring a DSB site near wild-type 70-bp repeat regions resulted in a minor cell cycle delay at 24 hours that was largely resolved by 48 hours ( Fig 4B—70-II-ISceI ) ., In contrast , deletion of the BES1 70-bp repeats resulted in a severe cell cycle defect that was only partially resolved at 48 hours post-induction ( Fig 4B—Δ70-ISceI ) ., To determine if the defect results from DSB formation , cell-cycle progression was monitored in the 70-bp repeat deletion mutant that does not harbor ISCEI ( Δ70-No ISCEI ) ., These cells did not have the cell cycle defect observed in ISCEI-expressing lines at 24 hours , but did display minor accumulation of cells in S-phase at 48 hours post-induction ( this could result from naturally occurring breaks arising late in growth that are not resolved normally in this line ) ., Together these data indicate that the growth delays observed in these cell lines are associated with cell cycle defects arising from DSB formation and suggest that the deletion of 70-bp repeats from the active site exacerbates these defects ., Multiple studies have shown that induction of ISceI-induced breaks in the active BES results in increased VSG switching , but the precise amount of switching can vary depending on the location of DSB formation an the activity of the ISCEI enzyme 12 , 14 ., To determine if the diversity of VSG gene selection ( reported in Fig 2 ) resulted from differences in switching dynamics , the VSG switching frequency was quantified for the ISceI-bearing cell lines and normalized to the number of population doublings ( Fig 4C , normalization derived from Fig 4A ) ., DNA break formation proximal to the long repeat region ( Fig 4C—70 . II ) resulted in approximately 100-fold increase in switching , compared to wild-type cells , as previously observed 12 ., DSB formation in the proximity of only three 70-bp repeats ( 70 . I ) resulted in a similar switching frequency , but a vastly different diversity in the selected VSGs ( 98% BES encoded VSGs selected compared with 52% in the 70 . II cell line Fig 2 ) ., This comparison underscores the role of the 70-bp repeat region in selection of VSGs from the genomic archive ., However , deletion of the 70-bp repeats resulted in a switching frequency , upon DSB induction , that was 10–100 fold greater than strains harboring 70-bp repeats ( Fig 4C—Δ70-ISceI ) , such that 1 in 10 cells had switched ( a frequency observable by flow-cytometry alone Fig 4D ) ., This was in contrast with the previous report of a similarly constructed strain 12 , probably because the slow growth phenotype had , in the previous study , led to the selection of a clone in which the ISceI site or enzymatic function was lost ., The switching frequency calculated after DSB formation in the 70-ISceI line is likely affected by the observed growth and cell cycle defects , so is not directly comparable to the values calculated for isogenic lines containing repeats , where DSB-induction does not noticeably affect growth and cell-cycle progression ., Nonetheless , the diminished capacity for VSG donor selection in the Δ70-ISceI line was definitely not the result of a reduction in switching frequency ., Based on the observation that the 70 . I-ISceI has a normal switching frequency and a modest capacity for archival VSG donor selection , we predicted that a minimal 70-bp repeat region could recapitulate the phenotypes associated with the long , cognate 70-bp repeat regions ., To test this prediction , the conserved 70-bp repeat sequence presented in Fig 1 was used to design synthetic 70-bp regions , which were introduced into the Δ70-ISceI landscape to produce stable cell lines and analyze their phenotypes ., The resulting cell lines , which harbor discrete repeat regions proximal to the ISceI site , are as follows: “Monomer” , which bears a single 70-bp repeat; “Dimer” , consisting of two monomeric units separated by a cognate spacer ( ATAATA ) ; and “Dimer_Rv” , which harbors the Dimer sequence in the opposite orientation with respect to transcription ( Fig 5A , repeat insertion sequences shown in S1 Text ) ., DSB induction in the Dimer cell line reduced the VSG switching frequency nearly 10-fold from Δ70-ISceI levels ( 2 . 3x10-2 compared with 1 . 9x10-1 , respectively ) , where strains harboring the 70-bp repeat Monomer or Dimer_Rv sequences were unchanged from the deletion mutant ( Fig 5B ) ., Similarly , the growth and cell cycle defects observed in the absence of 70-bp repeats ( Fig 4—Δ 70-ISceI ) were significantly improved by the addition of the Dimer region , while this was not the case for Monomer or Dimer_Rv lines ( Fig 5C and S3 Fig ) ., ( Phenotypes of an additional mutated repeat line “Mut_Dimer” did not suppress the Δ70-ISceI phenotypes shown only in S3 Fig ) ., If VSG switching and cell growth phenotypes correlate with VSG donor selection ( as suggested by data in Figs 2 and 3 ) , we would expect the Dimer cell line to result in selection of VSGs from within the genomic archive ., To determine the effect of the synthetic repeat sequences on VSG donor selection at an increased depth , RNA was extracted from DSB-induced post-MACS eluates from biological triplicates of these lines for VSG-seq analysis 28 ., The VSG-Seq method is distinct from the clonal analysis of VSG donor selection presented in Fig 2 in that it permits the identification of VSG RNAs comprising as little as 0 . 01% of the population 28 ., At this sensitivity , we observed that the line lacking repeats in the active BES ( Δ70-ISceI ) could occasionally select VSGs from sites other than BESs ( Fig 5D , supported by data in S5 Table ) , including two from metacyclic expression sites ( MES ) , one from a MC , and four from other undetermined ( UD ) loci ., Introduction of the repeat Dimer resulted in a significant ( pval = 0 . 0026 ) increase in the number of VSGs selected when compared with the no-repeat line ( average of 18 VSGs in Δ70-ISceI and 35 VSGs in Dimer populations , SI 9 ) ., This near doubling in the diversity of VSG selection was the result of a substantial increase in MC VSG selection ( pval = 0 . 005 , average Δ70-ISceI = 1 MC & Dimer = 11 MC ) and a more modest , but statistically significant , increase in the selection of VSGs arising from undetermined loci ( pval = 0 . 001 , average Δ70-ISceI = 4 UD & Dimer = 12 UD ) ., In contrast , addition of the Dimer_Rv sequence did not result in a significant increase in the selected VSG repertoire ( pval = 0 . 205 ) , although some subtle differences between Δ70-ISceI and Dimer_Rv strains can be observed ( Fig 5D ) ., These data have identified a minimal 70-bp repeat region able to partially suppress the collection phenotypes ( i . e . cell growth defect , cell cycle delay , increased VSG switching , and reduced VSG donor selection ) associated with DSB formation proximal to a 70-bp repeat deletion mutant and result in phenotypes similar to lines harboring cognate 70-bp repeats ., While unbalanced chromosomal translocations can fuel evolutionary change , they are generally deleterious to eukaryotic , especially mammalian , genomes ., African trypanosomes are a useful model of chromosomal translocations because their essential pathogenic process , antigenic variation , depends on them ., The early observation that genetic transposition of a new VSG into the active BES terminates within a tract of repetitive DNA inspired passionate functional speculation ., Yet , the available sequence information and genetic tools of the time ( and of studies that followed in the 1990s ) restricted the scope of possible analyses ., Thus , a viable hypothesis , that the repeats provide homology for recombination , became widely accepted 29 , 30 but was not tested ., In the present study we applied a variety of recently available sequencing databases ( BESs , trypanosome genomes , and VSGnome ) , a next-generation sequencing method ( VSG-seq ) , genetic tools ( including ISceI DSB induction ) , and cell biology assays ( such as VSG switching frequency quantification ) to test this long-standing hypothesis ., Classic sequencing approaches of the mid-1980s allowed three groups to determine the essential characteristics of the 70-bp repeats 17–19 and analysis of cosmid clones suggested that VSG genes and 70-bp repeats were widely distributed in the genome 31 ., Completion of the first African trypanosome genome sequencing project ( TREU927 ) confirmed , in detail , that the 70-bp repeat sequence is not only found at the BES subtelomeres but also proximal to VSGs on the chromosome arms 1 ., Yet , at that time , determining the degree of 70-bp repeat conservation within the genome was hindered by inherent challenges associated with assembling the sequences at the ends of chromosomes ., Here , we utilized existing comprehensive BES sequence data ( ABI 3730 , with approximately 700-bp read length 7 ) to produce a 70-bp consensus sequence and confirm its degree conservation among numerous BESs ., The length and conservation of this sequence corroborates some early findings 18 , but disagree somewhat with the often-asserted position that the 70-bp repeats are imperfect and have variable length 7 , 30 , 32 , 33 ., While the length of the AT-rich regions between conserved repeating units can vary , as reported 19 , we would suggest that the data presented in this study highlight the significance of the conserved repeating unit presented in Fig 1 ., It is important to note that the findings reported here do not address the putative function of the repetitive regions in DNA instability , the proposed function of the triplet repeats 19 , 34 ., The order and conservation of the 70-bp repeats inspired us to revisit the question of function ., Previous deletion of the BES1 70-bp repeat regions showed that the repeats themselves are not required for the low levels of gene conversion observed in vitro 20 ., This finding was significant in that it challenged long-held speculation that the repeats function as specific endonuclease-cleavage sites ., The recent availability of the Liste r427 VSGnome ( sequences of all VSG genes within the genomic archive ) 2 enabled testing of the second predicted function of the 70-bp repeats , namely providing homology in VSG donor selection ., However , the amount of switching that occurs in vitro is too low ( 1x10-6 ) to permit a substantive analysis of VSG switching outcomes ., This limitation was overcome through utilization of an artificial DNA breaking system that has been shown to increase the VSG switching frequency 12 , 14 , which occurs by gene conversion , in a similar manner to those that occur through more natural DNA break systems analyzed 13 ., Use of the established ISceI endonuclease cleavage system for DSB formation enabled in-depth analysis of how different regions , and mutations , of 70-bp repeats affect VSG switching and its outcomes ., The caveat , of course , is that ISceI is an artificial system and limits our interpretation of the implications for naturally occurring infections ., Nonetheless , this genetic tool enabled the observation of genetic phenomena that would not have been detectable otherwise ., Thus , individual clonal analysis of switched cells using the VSGnome resource allowed us to demonstrate that the BES encoded repetitive regions are required for selection of a normal repertoire of VSG genes , the first observed phenotype for the 70-bp repeats ., The increase in switching frequency following DNA break formation in the BES1 constructions presented here also enabled us to observe unexpected outcomes of 70-bp repeat deletion and variations ., Among these cell lines we observed that the 70-bp repeats have previously unappreciated and apparently connected effects on cell growth , cell cycle progression , and VSG switching following DSB formation ., The observation that deletion of 70-bp repeats results in significant cell cycle delays following DSB formation could suggest that , in comparison to lines harboring wild-type repeats , this cell line is defective for DNA break repair ., The fact that the same mutant cell line also switches much more frequently and results in increased cell death may suggest that cell lines harboring functional repeats process the DNA breaks more efficiently , as evidenced by the minimal cell cycle delay at 24 hours in 70 . II-ISceI ., These effects appear to depend on ISceI-induced DSB formation , as shown by the Δ70-No ISCEI cell line , whose behavior was largely unaffected by the repeat deletion , as expected from the literature 20 ., This collection of phenotypes was consistent among all cell lines that harbor “functional” ( 70 . II-ISceI , 70 . I-ISceI , and Dimer ) vs . “dysfunctional” ( Δ70-ISceI , Monomer , & Dimer_Rv ) 70-bp repeat regions ., Alternatively , similar phenotypes might be observed if the 70-bp repeats affect the ISceI cutting efficiency ., This could occur if there was steric hindrance at the cut site , which could result from binding proteins or DNA secondary structure ., Further exploration of the phenotypic alterations associated with the 70-bp repeat variations could lead to new mechanistic understanding of the requirements for the chromosomal translocations that support T . brucei antigenic variation ., Diverse pathogens utilize antigenic variation to escape the host immune system; among them T . brucei has the most extensive archive of surface antigen genes ( the VSGs ) 35 ., Yet , the extensive repertoire of VSG genes would be useless if they could not be activated ., Here we have shown that the 70-bp repeats are a key feature that permits access to the VSG archive ., This result experimentally validates previous speculations and extends our understanding by highlighting the specific DNA element , sequence , and orientation required for selection , at a depth of analysis only recently made possible by VSG-seq 28 ., It is unclear at this time if the 70-bp repeats influence the formation of new VSG gene variants through mosaicism , a known from of repertoire expansion mediated by recombination within VSG coding sequences ., At the sensitivity of VSG-seq and the discriminatory ability of its cognate assembly component , genes identified as VSG-variants ( Fig 5D—“var” ) appear to share similarities with mosaic VSGs ., BESs are essentially long homologous regions that not only contain the same organization , genes , and genetic elements , but are also nearly identical to one another at the sequence level for more than 50 kilobases 7 ., As homology length generally determines the frequency of recombination 36 , it is not unexpected that homologous BESs ( harboring many kb of 70-bp repeats ) are primary genomic sites favored during VSG selection ., What is surprising is the extent to which regions of wild-type repeats select sites other than BESs ( 48% of VSGs selected following induction of 70 . II-ISceI ) ., While VSG donor selection could be based on homology alone , our findings raise the possibility that another external factor , acting on the 70-bp repeats , promotes selection of non-BES encoded VSGs ., This effect could be in the form of repeat-specific DNA-binding proteins or be associated with subnuclear positioning during gene conversion , which is known to affect VSG expression 37 ., Overall , this study demonstrates that , following DSB formation and subsequent liberation of ssDNA , the 70-bp repeats guide homologous pairing toward diverse genomic sites that harbor the VSG archive , a repertoire-expansion function that is crucial to the long-term survival of the parasite in its host ., While the intricacies of VSG switching may be unique to African trypanosome parasites , the genetic processes described here have implications for chromosomal translocations that occur within other eukaryotic genomes ., The conserved repeat sequence was identified visually based on the BES sequences from T . brucei Lister 427 , and logos produced by http://weblogo . berkeley . edu/logo . cgi ., The same approach was then applied to TREU927 BES sequences ( S1 Fig and S1 Dataset ) ., The consensus sequence from the BES1 repeat logo was used to BLAST ( http://blast . ncbi . nlm . nih . gov/Blast . cgi ) the TREU927 genome and hits were called based on Max Score and Percent Identity ., VSG proximity was called based on the TREU927 genome annotation ., Contigs resulting from deep sequencing of the MC DNA fraction were used to determine repeat conservation and distance with respect to MC VSG genes 2 ., The sequence of each of the 11 megabase chromosomes from TREU927 ( T . brucei brucei ) , Lister427 ( T . brucei brucei ) , DAL972 ( T . brucei gambiense ) , IL3000 ( T . congolense ) , STIB805 ( T . evansi ) , and Y486 ( T . vivax ) genomes were downloaded from http://tritrypdb . org/tritrypdb/ and BLASTed ( http://blast . ncbi . nl | Introduction, Results, Discussion, Methods | African trypanosomes are mammalian pathogens that must regularly change their protein coat to survive in the host bloodstream ., Chronic trypanosome infections are potentiated by their ability to access a deep genomic repertoire of Variant Surface Glycoprotein ( VSG ) genes and switch from the expression of one VSG to another ., Switching VSG expression is largely based in DNA recombination events that result in chromosome translocations between an acceptor site , which houses the actively transcribed VSG , and a donor gene , drawn from an archive of more than 2 , 000 silent VSGs ., One element implicated in these duplicative gene conversion events is a DNA repeat of approximately 70 bp that is found in long regions within each BES and short iterations proximal to VSGs within the silent archive ., Early observations showing that 70-bp repeats can be recombination boundaries during VSG switching led to the prediction that VSG-proximal 70-bp repeats provide recombinatorial homology ., Yet , this long held assumption had not been tested and no specific function for the conserved 70-bp repeats had been demonstrated ., In the present study , the 70-bp repeats were genetically manipulated under conditions that induce gene conversion ., In this manner , we demonstrated that 70-bp repeats promote access to archival VSGs ., Synthetic repeat DNA sequences were then employed to identify the length , sequence , and directionality of repeat regions required for this activity ., In addition , manipulation of the 70-bp repeats allowed us to observe a link between VSG switching and the cell cycle that had not been appreciated ., Together these data provide definitive support for the long-standing hypothesis that 70-bp repeats provide recombinatorial homology during switching ., Yet , the fact that silent archival VSGs are selected under these conditions suggests the 70-bp repeats also direct DNA pairing and recombination machinery away from the closest homologs ( silent BESs ) and toward the rest of the archive . | Chromosomal translocations can fuel genetic change or cause catastrophic genomic damage ., African trypanosomes , exemplified by Trypanosoma brucei sub-species , are unicellular parasites that can chronically infect their human and livestock hosts by using a strategy of antigenic variation by which they repeatedly change their protein coats ., Switching the surface coat requires the accurate selection and translocation of a single silent coat gene , from a large genomic archive , into an actively transcribed site ., How the coat genes from within this deep archive are selected and activated was unproven ., Here we show that a specific repetitive DNA sequence is required to access coat genes from diverse sites within the genome ., The likely outcome of restricting this process of coat gene selection in natural infections would be a reduction in the chronic nature of African trypanosomiasis . | sequencing techniques, cell cycle and cell division, cell processes, cloning, parasitic protozoans, protozoans, archives, sequence motif analysis, molecular biology techniques, dna, dna recombination, gene conversion, research and analysis methods, research facilities, sequence analysis, genomics, chromosome biology, repeated sequences, molecular biology, information centers, biochemistry, trypanosoma, cell biology, nucleic acids, genetics, biology and life sciences, trypanosoma brucei gambiense, organisms, chromosomes | null |
journal.pgen.1002326 | 2,011 | Natural Selection Affects Multiple Aspects of Genetic Variation at Putatively Neutral Sites across the Human Genome | A substantial amount of effort in human population genetics has been aimed at understanding how natural selection operates in the human genome ., However , we lack a basic understanding of the importance of positive natural selection versus negative selection at shaping overall patterns of genome variation ., Thus far , most of the attention has been aimed at locating genes that have been under positive selection 1–19 ., These studies have identified several hundred candidates throughout the genome that may have been affected by positive natural selection ., However , fewer studies have attempted to gauge the prevalence of positive natural selection in the human genome ., Those that have attempted have come to very different conclusions ., Several studies suggested that positive selection may be common , with around 10% of the genome having been affected by a recent selective sweep 9 , 10 , 14 , 16 ., Other studies argued that selective sweeps were less common 20 , 21 ., Finally , some have estimated that approximately 10% , but perhaps up to 40% , of nonsynonymous human-chimp differences have been fixed by positive natural selection 22 , 23 ., Thus , there is little consensus regarding the importance of positive natural selection at shaping patterns of variability ., Additionally , the role of negative selection at shaping broad patterns of genetic variation across the genome needs to be clarified ., Many studies have suggested that nonsynonymous mutations and mutations in conserved noncoding sequences are weakly deleterious but may persist in the population due to genetic drift and other demographic phenomena 22 , 24–32 ., The effect that these weakly deleterious mutations have on nearby patterns of genetic variation remains unclear ., Furthermore , the importance of negative versus positive selection at shaping overall patterns of variation also remains ambiguous ., If natural selection ( either positive or negative ) is common in the genome , it should affect patterns of genetic variation at linked neutral sites across the genome 33 , 34 ., Selection may alter genetic variation in different ways ., We review these ways , discuss the empirical evidence for these effects , and highlight open questions that our study seeks to address ., First , selection may generate a correlation between levels of neutral diversity and recombination rate 35 , 36 ., This can occur under models with strong positive selection ( selective sweeps ) or negative selection acting on many deleterious mutations ( background selection ) ., Selective sweeps remove genetic diversity at linked neutral sites 33 , 37 ., In a region of the genome with a low recombination rate , a large length of sequence will have the same genealogy as the selected site ., As such , the selective sweep will remove neutral variation over a larger portion of the sequence in low recombination rate regions than in regions with higher recombination rates ., Background selection against deleterious mutations can also generate this correlation 34 , 38–41 ., Chromosomes carrying many deleterious mutations will be rapidly eliminated from the population ., Any neutral variation linked to the deleterious mutations will also be eliminated from the population ., This model predicts reduced variability in regions of the genome with low recombination rate because , as with the case of a selective sweep , a larger portion of the chromosome will share the same genealogy as the selected site ( s ) in regions of low recombination rather than in high recombination ., Several studies have searched for a correlation between diversity and recombination rate in humans ., Early studies based on a small number of genes came to conflicting conclusions ., Nachman et al . 42 , 43 found a significant correlation between diversity and recombination rate , but found no correlation between divergence and recombination rate , suggesting the effects of natural selection ., Hellmann et al . 44 , examining a different dataset , found that the correlation between diversity and recombination rate disappeared after correcting for human-chimp divergence ., They suggested that recombination may be mutagenic and that the original correlation was driven by co-variation of mutation and recombination rates ., Another study found that microsatellite diversity was not correlated with recombination rate 45 ., More recent studies on larger datasets have found significant correlations between diversity and recombination rate 46–48 ., These studies have found that the correlation between human diversity and recombination rate persists after controlling for human-chimp divergence ., While this is suggestive of the effects of natural selection , important features of this correlation have yet to be characterized ., For example , if natural selection is primarily driving the correlation , the correlation ought to be stronger in genic regions of the genome than in non-genic regions , because functional sites near genes are the most likely targets of selection ., This feature has yet to be explored ., Second , natural selection may generate a correlation between the allele frequency distribution and recombination rate ., Specifically , models of selective sweeps predict a skew toward an excess of low-frequency single nucleotide polymorphisms ( SNPs ) near the target of selection 49–51 ., Following the same logic as above , a larger region of the genome will be affected in areas with lower recombination rates , thus generating a correlation between allele frequency and recombination rate ., The effect of background selection on allele frequencies is less clear ., Simulation studies have suggested that intermediate strengths of background selection , especially in regions of low recombination , can generate a skew toward an excess of low-frequency SNPs 34 , 38 , 52–58 ., Most of the analytical formulae that describe background selection model the process as a reduction in effective population size , which does not predict a skew of the frequency spectrum ( 34 , 38–41 , but see Santiago and Caballero 59 ) ., Consequently , it has been argued that the effect of background selection on the frequency spectrum is rather weak , and as such , a skew toward low-frequency SNPs is more indicative of positive , rather than background selection 60–66 ., It is unclear whether there is a correlation between allele frequency and recombination rate in the human genome , though several small studies have found suggestive evidence 6 , 67 ., Furthermore , it is unclear which models of selection may be compatible with such a correlation ., Third , if selection is common , it ought to primarily affect patterns of genetic variation near genes because genes are the likely targets of selection ., Several studies have found that human-chimp divergence and human diversity were reduced near genes , suggesting the importance of selection at shaping overall patterns of variability throughout the genome 67–69 ., It is less clear whether there is a skew toward low-frequency alleles near genes ., Fourth , pervasive positive natural selection may generate a negative correlation between nonsynonymous divergence and levels of neutral genetic diversity ( 70–77 and reviewed in 78 ) ., The reason for this is that selective sweeps acting on amino acid changing mutations generate nonsynonymous fixed differences between species ., Regions of the genome that have been affected by these sweeps will likely also have reduced neutral polymorphism , thus generating the negative correlation between these two quantities ., It is unclear whether such a correlation can be generated in the absence of positive selection and how strong the correlation might be under various models of positive selection ., Here we further investigate these issues by studying patterns of genetic variation in three different genome-wide genetic variation datasets obtained from resequencing European individuals ., We find that levels of diversity are positively correlated with recombination rate and negatively correlated with genic content ., Minor allele frequency is also positively correlated with recombination rate and negatively correlated with genic content ., Using simulations , we show that these correlations are best explained by a model where many sites are under weak negative selection ., Models with numerous selective sweeps on nonsynonymous mutations predict too strong a negative correlation between neutral polymorphism and nonsynonymous divergence ., Though not required to explain the data , some smaller fraction of sites may be under positive selection ., Overall , this work points to the importance of weak negative selection at shaping patterns of variation throughout the human genome ., We analyzed genomic patterns of polymorphism from three genome resequencing datasets ., First , we analyzed low-coverage next-generation sequence data obtained from an exome-capture study of 2 , 000 Danish individuals ., Due to the non-specificity of the exome-capture arrays , portions of the genome outside of the targeted regions were sequenced , but at lower coverage ., Given the shallow sequencing depth across most of the genome ( roughly 0 . 1× per individual ) , it would be impossible to infer genotypes for each individual with any appreciable accuracy ., Instead , we implemented a statistical approach to estimate the population allele frequency of a SNP using the counts of different nucleotides at a particular site in the genome ( see Materials and Methods for a detailed description ) ., When combining reads across all individuals , approximately 30–40% of the genome had a sequencing depth of at least 100 reads ., We estimated the minor allele frequency ( MAF ) for all of these sites with a depth of at least 100 reads ., Those sites with an estimated MAF>5% were considered to be SNPs in this dataset and were used for subsequent analyses ., We used this conservative cut-off because of the difficulties in reliably estimating allele frequencies of rare alleles in low-coverage data 79 ., In order to verify patterns found in our low-coverage resequencing dataset , we also analyzed two other complementary datasets ., One dataset consisted of six European genomes that were sequenced to higher coverage ( denoted “higher coverage , ” see Materials and Methods for details ) ., The other dataset consisted of five genomes from Utah residents with ancestry from northern and western Europe ( abbreviated CEU ) and one genome from a Toscan individual sampled from Italy ( abbreviated TSI ) sequenced to high coverage by Complete Genomics ( denoted “CGS , ” see Materials and Methods ) ., Summaries of genetic variation were positively correlated across the three datasets ( Figure S1 and Figure S2 ) ., Due to the stochasticity of the evolutionary process , even with perfect data , patterns of polymorphism will not be perfectly correlated across different datasets ., To analyze correlations between different summaries of polymorphism and other genomic features , we divided the genome into non-overlapping 100 kb windows ( see Materials and Methods for further details ) ., Within each window , we tabulated the number of SNPs , average MAF , number of human-chimp differences , GC content , recombination rate ( as estimated from the high-resolution deCODE map 80 ) , fraction of each window where sequencing data was available , and the fraction of the window that overlaps with a RefSeq gene ., Since we wanted to examine the indirect effects of natural selection due to linkage , rather than assess the effects of natural selection on the selected sites themselves , all of our analyses removed the roughly 5% of the genome that was most conserved across species ( i . e . the phastCons regions 81 , see Materials and Methods ) ., These were the regions most likely to be directly under negative selection in the human genome 81 ., We then assumed that the remaining sequence that we analyzed was selectively neutral ., Because many of the genomic features were correlated with each other ( Table S1 , Table S2 , Table S3 ) , we performed partial correlation analyses to remove the effects of possible confounding variables ., The partial correlation can be thought of as the correlation between two variables when one or more other confounding variables are held constant ., We used partial correlations , rather than a full multivariate analysis , because the partial correlations have a simpler biological interpretation and have been used in other recent evolutionary studies 82 ., We found a strong positive correlation between the number of SNPs in a window and the recombination rate of the window ( Spearmans , Table S1 ) when looking at the low-coverage data ., We also observed a strong correlation between the number of human-chimp differences within a window ( d ) and recombination rate ( Spearmans , Table S1 ) ., When scaling diversity by divergence ( i . e . dividing the number of SNPs per covered base within a window by the number of human-chimp differences ) to potentially account for differences in mutation rate across the genome , we still found a strong correlation between scaled SNP diversity ( defined here as Snorm ) and recombination rate ( Spearmans , Table 1 , Table S1 ) ., In particular , regions of the genome with low rates of recombination ( i . e . <0 . 5 cM/Mb ) had especially low levels of polymorphism ., The rate of change of Snorm was less dramatic over the rest of the range of recombination rates ., We also found a positive correlation between Snorm and recombination rate when analyzing the higher-coverage and CGS datasets ( Spearmans , Table 1 , and Table S2 for the higher-coverage data; Spearmans , Table 1 , and Table S3 for the CGS data ) ., The correlation was even stronger than that observed in the low-coverage data ., We discuss several possible reasons for this difference in the Discussion section ., Nevertheless , the fact that we found the correlation in all three datasets strongly argues that it is a true biological correlation and not an artifact due to biases in the low-coverage Danish data ., The correlation between Snorm and recombination rate remained significant even after controlling for GC content , d , the number of neutral bases covered by sequencing data , and the fraction of genic bases within a window ( Table 1 ) , suggesting that these factors cannot completely explain this correlation ., Further , the average number of pairwise differences per window normalized by d was also positively correlated with recombination rate in both datasets ( Table S2 and Table S3 ) ., If natural selection is responsible for this correlation between Snorm and recombination rate , it may be stronger in genic regions of the genome than in non-genic regions ., The reason for this is that , all else being equal , genic regions will likely experience more natural selection than non-genic regions ., Non-genic windows were defined to be those that did not overlap with a RefSeq transcript ., Genic windows were those where at least half the window overlapped with a RefSeq transcript ., Indeed , the correlation was significantly stronger in genic windows than in non-genic windows in all three datasets ( P<0 . 0001 by permutation test , Figure 1 , Table 2 , Figure S3 , and Figure S4 ) ., This pattern holds even after controlling for confounding variables using a partial correlation analysis ., Inspection of the lowess lines in Figure 1A illustrates the differences between the correlation in genic and non-genic regions ., In genic regions with low recombination rates ( <0 . 5 cM/Mb ) , there is a sharp decrease in Snorm ., However , non-genic regions with low recombination rates did not show such a pronounced decrease in Snorm ( Figure 1A ) ., One concern with these analyses is that the low-coverage dataset was an exome resequencing dataset and the exome-capture process may have resulted in systematic differences between genic and nongenic regions ., However , we found the same pattern in the higher-coverage dataset and the CGS dataset , which were not targeted toward genes or exons ( Figure S3 and Figure S4 ) ., This argues that the differences between genic and non-genic regions were not due to systematic biases in the data , but rather to inherent differences between genic and non-genic regions of the genome ., We then examined the correlation between average MAF within a window and recombination rate ( Table 1 and Table S1 ) in the low-coverage data ., We found a weak , but statistically significant , positive correlation between these two variables ( Spearmans ) ., In regions of low recombination , there was a skew toward lower average MAF ., The correlation remained significant even after controlling for GC content , d , the number of neutral bases covered by sequencing data , and genic content , suggesting that it cannot be completely explained by these other factors ( Spearmans , Table 1 ) ., Finally , we also found a positive correlation between average MAF and recombination rate in the higher-coverage and the CGS data ( Table 1 , Table S2 , Table S3 ) , again suggesting that it was not due to biases in estimating SNP frequencies from low-coverage data ., A different summary of the frequency spectrum , Tajimas D 49 , also showed a correlation with recombination rate ( Table S2 and Table S3 ) , indicating that this correlation was not sensitive to the summary of the frequency spectrum employed ., However , no clear pattern emerged when testing whether the correlation between average MAF and recombination rate was stronger in genic versus non-genic regions ., For all three datasets , the pairwise correlation between average MAF and recombination rate was higher in genic regions than non-genic regions ( P<0 . 05 , by permutation test , Figure 1B , Figure S3B , Figure S4B , Table 2 ) ., In the higher-coverage dataset , genic regions showed a stronger correlation between MAF and recombination rate than non-genic regions even after controlling for GC content , d , and the number of bases covered by sequencing data using a partial correlation analysis ( P<0 . 02 by permutation test , Table 2 ) ., However , after controlling for the confounding variables , there was little difference in the partial correlation coefficients between genic and non-genic regions in the low-coverage and the CGS datasets ( Table 2 ) ., Thus , there was no clear evidence suggesting that the correlation between MAF and recombination rate was stronger in genic than non-genic regions of the genome ., This may not be surprising because this correlation was quite weak , making it difficult to detect subtle changes in its strength across the genome ., If natural selection affects patterns of genetic variation across the genome , Snorm , average MAF , and d may be reduced in windows of the genome that contain more genic bases ., These patterns would be expected if most of the selection in the genome occurs near genes , rather than in intergenic regions ., Indeed , in all three datasets , we found a negative correlation between Snorm and the fraction of bases within a window that overlapped with a RefSeq transcript ( Table 1 ) ., In other words , windows with a higher genic content tended to have fewer SNPs ., These correlations became stronger when controlling for d , recombination rate , the fraction of the window with sequencing coverage , and GC content ( Table 1 ) ., There was a weak , but significant , negative correlation between MAF and fraction of bases that overlapped with a RefSeq transcript in all three datasets examined ( Table 1 ) ., Windows with a higher genic content tended to have lower average MAF than windows with lower genic content ., In the low-coverage and higher-coverage datasets , the correlation became stronger when controlling for d , recombination rate , the fraction of the window with sequencing coverage , and GC content ( Table 1 ) ., Finally , we found a very strong negative correlation between d and the fraction of genic bases within a window ( Spearmans , Table S1 , Table S2 , Table S3 ) ., These results were in agreement with those from a study 67 which found reduced diversity and divergence near genes even after removing the regions of the genome most conserved across species ( i . e . the phastCons elements ) ., We next tested whether there was a correlation between Snorm and the number of nonsynonymous human-chimp differences within a window ( DN ) ., A negative correlation between these two variables has been interpreted as evidence of selective sweeps across the genome ( 70–77 and reviewed in 78 ) ., When tabulating DN , we did not remove sites which were conserved across species ., We observed weak negative correlations between Snorm and DN as well as between Snorm and the number of synonymous human-chimp differences ( DS ) for several of the datasets ( Table S4 ) ., However , when we normalized DN by the number of nonsynonymous sites per window ( the normalized value is called dN ) or used a partial correlation analysis to control for the number of nonsynonymous sites per window , none of the datasets showed a significant negative correlation ( Table S4 ) ., The same was true for synonymous human-chimp differences ., Haddrill et al . 77 suggested that a negative correlation between Snorm and dN may be more apparent in genes with elevated dN ., Thus , we also tested for a correlation between Snorm and dN using only the windows in the 90th percentile of dN ., In general , the values of Spearmans were more negative in this subset of the data than when analyzing the entire dataset ( Table S5 ) ., For example , in the CGS data , when controlling for d , GC content , recombination rate , the number of nonsynonymous sites , and the fraction of the window with sequencing coverage ., However , Snorm was also negatively correlated with dS in the windows in the 90th percentile of dS ( , controlling for d , GC content , recombination rate , the number of synonymous sites , and the fraction of the window with sequencing coverage ) ., The fact dS showed a similar negative correlation with Snorm as dN did , combined with the fact that synonymous sites are usually assumed to be neutrally evolving in humans , suggested that these correlations may have been driven by a neutral process , rather than positive selection ., One possibility was that the recent fixations of neutral synonymous or nonsynonymous mutations led to a decrease in neutral diversity , as suggested by earlier theoretical work 83 ., As such , regions with high dN ( or high dS ) would have lower Snorm , generating the negative correlation ., Overall , these results suggest that regions of the genome that have more nonsynonymous human-chimp differences do not have lower levels of neutral polymorphism , beyond the reduction in diversity already expected in genic regions of the genome or surrounding neutral fixations ., We next evaluated whether population genetic models including population size changes , recombination rate variation , and natural selection could generate the correlations that we observed in the empirical datasets ., We simulated 100 kb regions consisting of exons , introns , and an intergenic sequence ( see Materials and Methods , Figure S5 ) ., We examined several different models of selection ( see Table S6 for the specific parameter values ) and examined the correlation between patterns of genetic variation in the neutrally evolving intergenic sequence and other genomic attributes ., Because many studies have found that nonsynonymous mutations are weakly deleterious 22 , 26 , 28 , 84 , one model included weak negative selection acting only on nonsynonymous sites ( shown in purple in Figure 2 ) ., It had been suggested that conserved noncoding sites are also likely to be weakly deleterious 25 , 27 , 31 , so another model included negative selection acting on a fraction of intronic sites ( shown in blue in Figure 2 ) ., In the third model ( shown in orange in Figure 2 ) , most mutations at nonsynonymous positions were negatively selected , but a small fraction was positively selected ., Finally , the fourth model added weak negative selection at a fraction of intronic sites to a model where most mutations at nonsynonymous positions were negatively selected , but a small fraction was positively selected ., Our simulations confirmed previous predictions that both hitchhiking and background selection 33 , 34 , 37–41 could generate a positive correlation between genetic diversity at linked neutral sites and recombination rate ( Figure 2A and Figure S6A ) ., Importantly , these simulations demonstrated that the background selection effect can occur with weak negative selection acting on many sites simultaneously ., Models with negative selection acting on noncoding and coding mutations , as well as models with positive selection , could generate positive correlations similar to those in the observed data ( red lines in Figure 2A and Figure S6A ) ., Models of natural selection predicted a positive correlation between average MAF at linked neutral sites and recombination rate ( Figure 2B and Figure S6B ) ., The strongest correlations seen for models with only negative selection were for intermediate strengths of selection ( e . g . 25% of intronic sites with s\u200a=\u200a2 . 5×10−4 ) ., Stronger selection ( s\u200a=\u200a5×10−3 ) resulted in a weaker correlation ( Table S7 ) ., Importantly , models that contained no sites under positive selection predicted a correlation between MAF and recombination rate roughly similar in magnitude to that seen in the observed data ( red lines in Figure 2B and Figure S6B ) ., These results suggest that both positive and weak negative selection were capable of affecting allele frequencies at linked neutral sites ., Thus , a correlation between allele frequency and recombination rate cannot be taken as unambiguous evidence of positive selection ., In some cases , the correlation coefficients between MAF and recombination rate and diversity and recombination rate were significantly higher than zero under purely neutral models ( Figure 2 and Figure S6 ) ., We performed coalescent simulations using ms 85 under the standard neutral model with different rates of recombination to further investigate this issue ., Not only was the variance of the distribution of diversity ( or average MAF ) greater in simulations without recombination , but the shape of the distribution changed depending on the recombination rate ., For example , in the case of a high recombination rate , the distribution of the number of segregating sites approached a Poisson distribution , and was symmetric about its mean ., However , with no recombination , the distribution became less symmetric , with a higher mass below the mean and a longer tail to the right ( Figure S7 ) ., Thus , the median of the distribution of diversity simulated with no recombination was lower than the median of the distribution with the high recombination rate ., As such , a weak positive correlation between recombination rate and diversity may be expected ., The same arguments hold for understanding the correlation between MAF and recombination rate ( Figure S7 ) and Tajimas D and recombination rate ( Figure S7 , see also 63 , 86 ) ., Since we used simulations to interpret the correlations observed in the actual data , this effect did not alter our interpretation ., Previous authors ( 70–77 and reviewed in 78 ) had suggested that a negative correlation between neutral polymorphism and nonsynonymous divergence may be a signature of positive selection that cannot be generated by negative selection and/or demographic processes ., In our simulations , a model with negative selection acting on noncoding sites , but where a fraction of coding mutations were positively selected showed a negative correlation between Snorm and dN ( orange points in Figure 2C and Figure S6C ) ., Models that did not include any positive selection , but included negative selection on a fraction of noncoding sites ( blue points in Figure 2C and Figure S6C ) , showed little correlation between these two variables ., Thus , for the models investigated here , the negative correlation was specific to models of positive selection ., As such , it may offer a way to distinguish between models of negative and positive selection ., However , a significant negative correlation was not always seen in models that included some sites under positive selection ( green points in Figure 2C and Figure S6C ) ., Instead , the correlation was influenced by the relative amounts of negative versus positive selection ., Negative selection made the correlation more positive , while positive selection made the correlation more negative ., The correlation ultimately observed was due to the net effect of both types of selection ., We next used the simulations to evaluate what role positive selection may have played in shaping patterns of variability across the genome ., We first examined models with only strong positive selection ., A model where 0 . 5% of nonsynonymous mutations were positively selected ( s\u200a=\u200a0 . 625% ) could generate the observed correlation between Snorm and recombination rate ( black , p+\u200a=\u200a100% , p−\u200a=\u200a0% in Figure 3A; p+ denotes the proportion of simulated windows where positive selection could occur ) ., However , this model predicted too strong a negative correlation between Snorm and dN to be compatible with the data ( black , p+\u200a=\u200a100% , p−\u200a=\u200a0% in Figure 3B ) ., Because several studies have suggested that 0–10% of the genome has been affected by a selective sweep 9 , 10 , 14 , 16 , 20 , 21 , we next examined a model where 5% of the simulated windows included positive selection ., A model where the remaining 95% of the windows were neutral does not predict a correlation between Snorm and recombination rate strong enough to match the actual data ( black , p+\u200a=\u200a5% , p−\u200a=\u200a0% in Figure 3A ) ., This suggests that a small number of positively selected sites by themselves are not sufficient to generate this correlation ., Further , this model still predicted a negative correlation between Snorm and dN ( black , p+\u200a=\u200a5% , p−\u200a=\u200a0% in Figure 3B ) ., However , a model where 5% of the simulated windows included positive selection and the remaining 95% of windows included negative selection on coding and noncoding sites predicted a correlation between Snorm and recombination rate similar to that observed in the actual data ( black , p+\u200a=\u200a5% , p−\u200a=\u200a95% in Figure 3A ) ., Because adding negative selection resulted in an increase in the strength of this correlation , we concluded that the correlation observed in the data has been primarily driven by negative selection ., Also , under this model , the negative correlation between Snorm and dN was very weak and was compatible with that from the actual data ( black , p+\u200a=\u200a5% , p−\u200a=\u200a95% in Figure 3B ) , presumably because most of the windows have been subjected to negative selection ., A model where the strength of positive selection was weaker showed similar trends ( pink points in Figure 3 ) ., This analysis indicated that the correlation between neutral diversity and recombination rate was primarily driven by many weakly deleterious polymorphisms across the genome , rather than by a small proportion of strongly positively selected mutations ., Finally , our simulations ( Figure 4 ) suggest that negative or positive selection can generate a strong correlation between neutral human-chimp divergence ( d ) and recombination rate even when the mutation rate is constant across all simulation replicates ., This correlation was likely driven by selection occurring in the ancestral population 67 , 87 ., Thus , the correlation between d and recombination rate can be readily explained by mechanisms other than recombination itself being mutagenic 44 , 46 , 88 ., We have examined patterns of putatively neutral g | Introduction, Results, Discussion, Materials and Methods | A major question in evolutionary biology is how natural selection has shaped patterns of genetic variation across the human genome ., Previous work has documented a reduction in genetic diversity in regions of the genome with low recombination rates ., However , it is unclear whether other summaries of genetic variation , like allele frequencies , are also correlated with recombination rate and whether these correlations can be explained solely by negative selection against deleterious mutations or whether positive selection acting on favorable alleles is also required ., Here we attempt to address these questions by analyzing three different genome-wide resequencing datasets from European individuals ., We document several significant correlations between different genomic features ., In particular , we find that average minor allele frequency and diversity are reduced in regions of low recombination and that human diversity , human-chimp divergence , and average minor allele frequency are reduced near genes ., Population genetic simulations show that either positive natural selection acting on favorable mutations or negative natural selection acting against deleterious mutations can explain these correlations ., However , models with strong positive selection on nonsynonymous mutations and little negative selection predict a stronger negative correlation between neutral diversity and nonsynonymous divergence than observed in the actual data , supporting the importance of negative , rather than positive , selection throughout the genome ., Further , we show that the widespread presence of weakly deleterious alleles , rather than a small number of strongly positively selected mutations , is responsible for the correlation between neutral genetic diversity and recombination rate ., This work suggests that natural selection has affected multiple aspects of linked neutral variation throughout the human genome and that positive selection is not required to explain these observations . | While researchers have identified candidate genes that have evolved under positive Darwinian natural selection , less is known about how much of the human genome has been affected by natural selection or whether positive selection has had a greater role at shaping patterns of variation across the human genome than negative selection acting against deleterious mutations ., To address these questions , we have combined patterns of genetic variation in three genome-wide resequencing datasets with population genetic models of natural selection ., We find that genetic diversity and average minor allele frequency are reduced in regions of the genome with low recombination rate ., Additionally , genetic diversity , human-chimp divergence , and average minor allele frequency have been reduced near genes ., Overall , while we cannot exclude positive selection at a fraction of mutations , models that include many weakly deleterious mutations throughout the human genome better explain multiple aspects of the genome-wide resequencing data ., This work points to negative selection as an important force for shaping patterns of variation and suggests that there are many weakly deleterious mutations at both coding and noncoding sites throughout the human genome ., Understanding such mutations will be important for learning about human evolution and the genetic basis of common disease . | neutral theory, population genetics, evolutionary selection, mutation, effective population size, genetic polymorphism, comparative genomics, biology, evolutionary theory, evolutionary genetics, genetic drift, adaptation, natural selection, genetics, evolutionary biology, evolutionary processes, genetics and genomics | null |
journal.pntd.0000236 | 2,008 | The Effect of Azithromycin on Ivermectin Pharmacokinetics—A Population Pharmacokinetic Model Analysis | The operational efficiency of disease elimination programs in developing countries could be improved by integrating delivery of several interventions at local ( village and district ) levels 1–3 ., In areas endemic for co-infection with filarial nematodes and Chlamydia trachomatis , one such integrated disease elimination strategy would be based on mass administration of a three-drug combination: ivermectin for onchocerciasis , albendazole for lymphatic filariasis and azithromycin for trachoma ., Regular administration of this combination would also be predicted to reduce other infectious agents including soil transmitted nematodes and bacterial sexually transmitted diseases 4 ., A recent pharmacokinetic study evaluated co-administration of azithromycin , ivermectin and albendazole 5 , and showed that mean ivermectin pharmacokinetic parameters , area under the concentration-time curve ( AUC ) and maximum concentration ( Cmax ) , were increased by 31% and 27% , respectively relative to a baseline period ., The variability in this interaction was large , with two individuals having 3-fold increases in ivermectin AUC ., Increased ivermectin exposures could potentially have safety implications , as high dose ivermectin animal studies and observations of human overdose have reported signs and symptoms of central nervous system ( CNS ) toxicity including emesis , mydriasis and ataxia 6 ., However a recent safety study demonstrated no significant toxicity in the CNS or other body systems , with ivermectin doses up to 10 times the highest labeled dose of 200 µg/kg 7 , 8 ., The purpose of this analysis was to model the ivermectin pharmacokinetic data from the recently reported interaction study 5 , to further characterize the interaction , and explore the sources of variabilities between subjects and across treatments ., The model was also used to simulate the outcomes of 1000 trials , to ensure that peak ivermectin exposures seen during co-administration did not exceed those observed in the high dose safety and pharmacokinetic study 7 ., Data from a historical Phase I study with intensive sampling in healthy subjects was used to develop a population pharmacokinetic model for ivermectin 5 ., All subjects provided written informed consent according to local requirements before entering the study , and the protocol and Informed Consent Form were approved by the local Institutional Review Board ., This was a three-arm crossover study , where subjects were administered single-dose regimens of the following treatments in random order:, ( i ) azithromycin 500 mg;, ( ii ) ivermectin 200 µg/kg of total body weight rounded to the nearest 3 mg plus albendazole 400 mg; and ,, ( iii ) all 3 drugs administered concurrently ., All doses were administered with 240 mL of water and a standardized breakfast ., Prior to dosing and breakfast , subjects fasted overnight and then abstained from any further food for 4 hours after study drug administration ., Study arms were separated by washout periods of 3 weeks ., Full details of the study are provided in 5 ., Blood samples were collected predose and at 0 . 5 , 1 , 1 . 5 , 2 , 3 , 4 , 6 , 8 , 10 , 12 , 24 , 36 , 48 , 72 , 96 , 120 , 144 , and 168 hours after drug administration during each of the study phases ., Samples were collected into heparinized Vacutainers ., Blood samples were centrifuged at 3000 rpm for 15 minutes and the plasma samples were collected in plain plastic tubes without anticoagulant and then stored at −80°C ., Samples were shipped frozen overnight on dry ice to BAS Analytics ( West Lafayette , IN ) for sample analyses ., Ivermectin is detected in the body as two metabolites ( 22 , 23-dihydroavermectin-B1a ( H2B1a ) and 22 , 23-dihydroavermectin-B1b ( H2B1b ) , and these were assayed using a validated high performance liquid chromatography system with liquid chromatography/mass spectrographic detection ., The assays were linear over the ranges of 2 . 5–1000 . 0 ng/mL and 2 . 5–20 . 0 ng/mL , respectively ., The precision values for both assays were <10% ., In terms of accuracy , while the bias was not exceeded ( ±15% ) for H2B1b for either the high or low quality control ( QC ) samples , they were for H2B1a during long-term stability testing ( −21 . 8% at the low QC and −17 . 3% for the high QC ) ( see 5 ) ., Plasma concentration-time data were analyzed using standard noncompartmental analytical software ( WinNonlin 4 . 1; Pharsight Corporation , Mountain View , CA ) , and key parameters are shown in Figure 1 ., The data analysis presented here is for ivermectin data from the ivermectin plus albendazole arm ( Baseline Phase ) , and from the ivermectin , albendazole plus azithromycin arm ( Interaction Phase ) ., Eighteen healthy Caucasian volunteers were enrolled in and completed this study ( 9 males and 9 females , mean ±SD age , 39 . 4±10 . 5 years , weight 78 . 2±12 . 4 kg , ivermectin dose 15 . 5±2 . 6 mg ) ., All the data from both arms of the cross-over study were fitted simultaneously ., The data set contained pooled pharmacokinetic , demographic/covariate , and dosing information ., Data were analyzed using nonlinear mixed-effects modeling with the NONMEM software system , Version V , Level 1 . 1 ( GloboMax LLC , Ellicott City , MD ) with the PREDPP model library and NMTRAN subroutines ., Computer resources included personal computers with Intel Pentium 4 processors , Windows XP Professional operating system , the GNU Fortran Compiler , GCC-2 . 95 ( Win-32 version also known as G77; GNU Project , http://www . GNU . org/ ) ., Key pharmacokinetic parameters from the modeling are described in Figure 1 ., The first-order conditional estimation method with η-ε interaction ( FOCEI ) was employed for all model runs ., Individual estimates of pharmacokinetic parameters were obtained using POSTHOC ( an empirical Bayesian estimation method ) ., The random effect models sufficiently described the error distributions ., For this analysis all interindividual errors were described by exponential error models on selected parameters ( Equation 1 ) ., ( 1 ) where: Pi is the true parameter value for individual i , is the typical population value ( geometric mean ) of the parameter , ηPi are individual-specific interindividual random effects for individual i and parameter P and were assumed to be independently and identically distributed following a normal distribution with mean 0 and variance omega ( ω ) squared ( η∼N ( 0 , ω2 ) ) ., The data could not support a full covariance block for the OMEGA matrix ., Modeling began with the assumption of no covariance between interindividual random effects ( diagonal ω matrix ) ., Later , the covariance between clearance ( CL ) and volume of distribution in the central compartment ( Vc ) was estimated ., For pharmacokinetic observations in this analysis , the residual error model was described by a combined additive and proportional error model ( Equation 2 ) ., ( 2 ) where: Cij is the jth measured observation ( plasma concentration ) in individual i , is the jth model predicted value ( plasma concentration ) in individual i , εpij and εaij are proportional and additive residual random errors , respectively , for individual i and measurement j and are assumed to be independently and identically normally distributed , following a normal distribution with mean 0 and variance sigma ( σ ) squared ( ε∼N ( 0 , σ2 ) ) ., For each treatment arm , separate residual errors were explored ., The pharmacokinetic models were evaluated for goodness of fit and were then subjected to predictive check model evaluation ., For more detailed technical information on these methods , please see NONMEM users guide 9 ., After the structural pharmacokinetic model was established , known physiologic relationships were incorporated into the covariate-parameter models ., For example , the change in physiologic parameters as a function of body size was both theoretically and empirically described by an allometric model ( Equation 3 ) 10 ( 3 ) where: the typical individual value of a model parameter ( TVP ) was described as a function of individual body weight ( WTi ) , normalized by a reference weight ( WTref ) , which was 70 kg ., θTVP is an estimated parameter describing the typical pharmacokinetic parameter value for an individual with weight equal to the reference weight and θallo is an allometric power parameter ( which can be estimated or fixed to a value of 0 . 75 for clearances , and a value of 1 for anatomical volumes ) ., Assessment of model adequacy and decisions about increasing model complexity were driven by the data and guided by goodness-of-fit criteria , including:, ( i ) visual inspection of diagnostic scatter plots ( observed vs . predicted concentration , residual/weighted residual vs . predicted concentration or time , and histograms of individual random effects;, ( ii ) successful convergence of the minimization routine with at least 2 significant digits in parameter estimates;, ( iii ) plausibility of parameter estimates;, ( iv ) precision of parameter estimates;, ( v ) correlation between model parameter estimation errors <0 . 95 , and, ( vi ) the Akaike Information Criterion ( AIC ) , given the minimum objective function ( OBJ ) value and number of estimated parameters 9 ., The criteria for successful runs were restricted to successful convergence using FOCE with interaction , good diagnostics for the model-fit for all data of the different treatment periods , and reasonable estimates for fixed and random effect parameters ., Model evaluations included comparisons of the OBJ between hierarchical models ., A decrease in OBJ corresponding to a chi-square distribution with α\u200a=\u200a0 . 01 and degrees of freedom equal to the difference in the number of estimated parameters between the two models was used as the criterion for model comparisons ., Final model parameter estimates were reported with a measure of estimation uncertainty including the asymptotic standard errors ( obtained from the NONMEM $COVARIANCE step ) ., A limited covariate modeling approach emphasizing parameter estimation given the available data , rather than stepwise hypothesis testing , was implemented for this population pharmacokinetic analysis ., The study population contained equal numbers of males and females ., As such , age , weight and gender were explored as potential covariates ., First , pre-defined covariate-parameter relationships were identified based on exploratory graphics , mechanistic plausibility of prior knowledge , and then a full model was constructed , with a fixed allometric relationship of body weight on clearance and volume parameters ., Interindividual variability could not be incorporated on all fixed-effects parameters to get successful FOCE runs ., For residual variance , a separate residual error was assigned for each of the treatment arms ., A combined additive and proportional error model was used with 4 parameters to be estimated for the residual error ., Various population models were evaluated , but only two models that best described the data ( as determined by the log likelihood criterion and visual inspection ) are presented ., The first modeling approach was a population model that included all subjects ., Because some of the modeling parameters and their variances were clearly not normally distributed , and showed asymmetric distribution , a mixture model was developed ., A second modeling approach was a population mixture model as it met our criteria for model adequacy and provided supporting evidence of the dichotomy of the observed individual data ., Each subpopulation would have an associated submodel with different fixed or random effects ., This model was adopted to accommodate the fact that only some of the individuals exhibited a pronounced increase in ivermectin bioavailability during the interaction arm of the study ., It was preferred over a population model with and without outlier individuals , as it gave a better fit to the data as measured by change in OBJ , and met our criteria for a successful run in terms of a complete successful convergence with reasonable estimate for precision for both fixed and random effects ., Model development was guided by various goodness-of-fit criteria , including diagnostic scatter plots ., Checking of the individual fits was also employed as part of judging the model performance for each patient ., The final model and parameter estimates were then investigated with the predictive check method ., This method was similar to the previously described posterior predictive check , but assumes that parameter uncertainty is negligible , relative to interindividual and residual variance 11 ., The basic premise is that a model and parameters derived from an observed data set should produce simulated data that are similar to the original observed data ., The predictive check is a useful adjunct to typical diagnostic plots , in that the predictive check provides information about the performance of random-effects parameter estimates , whereas typical diagnostic plots are primarily informative about the fixed-effects parameter estimates ., The predictive check model evaluation step was performed by using the final model and its parameter estimates to simulate data under the same experimental design of the original data ., One thousand Monte Carlo simulation replicates of the original data set were generated using the final non-mixture and mixture population pharmacokinetic models ., Distributions of Cmax across all data simulations were compared with Cmax distribution in the observed data set ., The simulated data from each of the 1000 virtual trials ( 18000 subjects for each treatment period ) were assembled , and the similarity between the actual observed data and simulated data was examined by comparing the 95% predictions intervals of the simulated data with the original observed data ., Ivermectin concentration-time data were best described by a two-compartment pharmacokinetic model with first-order elimination and absorption ( Figure 3 ) ., Covariance between CL and Vc elements of the OMEGA matrix was incorporated in the model ., The use of different residual variance models stratified by the treatment with and without shared additive components was explored and incorporated into the structural model ., Inclusion of age or gender as covariates did not contribute additional information for explaining pharmacokinetic variability based on OBJ differences in hierarchical models , model convergence , as well as diagnostic graphics ., Therefore , none of these covariates was included as a covariate in the final population pharmacokinetic model ., Importantly , the available data for this investigation contained a relatively small number of subjects and a limited age range , and so formal hypothesis ( significance ) testing for covariate effects was not considered ., The final non-mixture model had 7 fixed-effect parameters and 8 random-effect parameters as shown in Table, 1 . Population pharmacokinetic parameters ( CL , Vc , Q , Vp; see Figure 1 ) were standardized to a 70 kg person using the allometric size model 10 ., In parametric nonlinear mixed effects modeling , the distribution of ηs is assumed to be normal ( mean\u200a=\u200a0 , variance\u200a=\u200aω2 ) ., With each model developed , we checked the distribution of ηs , and their mean values ., The η distribution indicated a clear violation of the normality assumption ., It was necessary to modify the original model to improve η distribution diagnostics ., A mixture modeling approach was considered as the distribution of some of the pharmacokinetic parameters and inter-individual variabilities indicated a lack of homogeneity ., The final mixture model had 9 fixed-effect parameters and 8 random-effect parameters as shown in Table, 2 . Goodness-of-fit plots for the final model are shown in Figure 4 ., The mixture model differed from the non-mixture model in only two parameters: one defining the difference between the two subpopulation in terms of bioavailability , the second defining the partition of the population between the two subpopulations ., Using this approach , inter-individual variability distribution was modeled as two subpopulations ( A and B ) ., The unknown mixture distribution was estimated at an individual level ., The estimate for each subpopulation included different fixed effects parameters , different variance parameters , estimation of fraction of individuals in each subpopulation , and each individual was assigned to the most likely subpopulation ., The proportion of subjects in subpopulations A and B was estimated as 55% and 45% , respectively ., Both the population and individual predictions adequately described the AUC profiles for each subject ( Figure 5 ) , as displayed by the baseline and interaction phases for subpopulation B . A similar fit of individual data was observed for Subpopulation A ( data not shown ) ., Figure 6 displays median , 97 . 5th , and 2 . 5th quantiles of the simulated data as lines with the observed data plotted as individual points ., Less than 5% of the observed data were outside these 95% prediction intervals ., No biased pattern or any tendency for over- or underestimation was noted for the different treatment periods , or for the two subpopulations ., This finding gives confidence in the model performance in predicting the expected ivermectin exposures under different circumstances ., Simulated maximum concentrations for each individuals Cmax values were summarized across 1000 simulation replicates of the original population pharmacokinetic database and plotted as box plots ( Figure 7 ) ., The upper panel shows box plots of the observed ivermectin Cmax for baseline and interaction periods for all subjects , and for the two subpopulations ., The lower panel shows box plots for ivermectin Cmax from 1000 simulated trials for the non-mixture model ( all subjects ) , and the mixture model ( subpopulations A and B ) ., The mixture model pattern predictions for the two subpopulations were very consistent with the observed data 5 ., Extreme values were: non-mixture model: 201 . 2 ng/mL; mixture model subpopulation A: 115 . 3 ng/mL; B: 175 . 5 ng/mL ., There are a number of interesting findings from this analysis of data from an interaction study of ivermectin and azithromycin ., This is the first published population model of ivermectin pharmacokinetics ., It demonstrates the utility of population mixture modeling as an approach to explore drug interactions , especially where there may be population heterogeneity ., The mechanism for the interaction was identified ( an increase in bioavailability in one subpopulation ) ., The model was used to simulate multiple clinical trials , to identify the maximum exposures that might be observed during co-administration , which permits comparison with previously published safety and pharmacokinetic data ., Ivermectin has been approved for use in humans for 2 decades , yet relatively limited pharmacokinetic data have been published ., Recent studies using modern assay methods have characterized its pharmacokinetics using noncompartmental methods in the context of drug combination studies for treatment of onchocerciasis and lymphatic filariasis 12–14 , or in high doses for treatment of head lice 7 ., The calculated model parameters are in close agreement with those determined using noncompartmental methods 5 ., A two compartment model is consistent with the disposition of ivermectin in man and other species , with a high volume of distribution into a peripheral compartment 15 ., Ivermectin is metabolized extensively in the liver via cytochrome P450 isozyme ( CYP ) 3A4 16 ., It is both a substrate for the transporter P-glycoprotein ( Pgp ) 17 , 18 , as well as a moderately potent Pgp inhibitor at concentrations consistent with clinical exposures in the present study ( IC50 0 . 18–0 . 4 µM; 19 , 20 ) ., The variability of the magnitude of change in ivermectin pharmacokinetics observed in the interaction phase 5 complicated the interpretation of the presence or absence of a drug interaction , as the response was very inconsistent among individuals ., One of the objectives of this analysis was to explore how nonlinear mixed-effects modeling could be used to analyze such heterogeneous and highly variable experimental data from a relatively small number of subjects , with intensive pharmacokinetic sampling ., The initial non-mixture model provided an adequate description of ivermectin pharmacokinetic data , however interindividual variability was not homogeneous and could not be explained by the available covariates ., A mixture model was able to resolve this , and provided an explanation for the observed differences in bioavailability seen in the clinical study ., Mixture modeling assumes two or more subpopulations exist , rather than a single homogeneous one 21 , and the final model has two additional fixed parameters , one relating to subpopulation differences in ivermectin bioavailability , and the other defining the two subpopulations ., The final mixture model provided a good description of ivermectin data from both treatment periods ., Goodness-of-fit criteria revealed that the final model was consistent with the observed data and that no systematic bias remained ., The data points ( Figure 4 ) are scattered closely and randomly around the line of identity , and the homogenous and random distributions of weighted residuals indicate the error model was suitable for describing the variance of the data ., The model evaluation results provided evidence that both the fixed-effects and random-effects components of the final model were reflective of the observed data as well ., The fact that less than 5% of the data were located outside the 2 . 5-97 . 5th quantile range suggests that the model accurately describes the central tendency and the variability of the data for the two subpopulations and for the two treatment periods , despite the large number of parameters and the low number of patients who participated in the study ., The predictive check shows there is no bias at any phase of the pharmacokinetic profile , which makes the model useful in predicting ivermectin blood concentrations , when given alone or co-administered with azithromycin ., Typically , a mixture modeling approach would not be considered at the outset of a population pharmacokinetic analysis ., Because of the unexplained remaining variability ( see above ) , in the present analysis , the following decision rules were used in the evaluation of the mixture model:, ( i ) The Estimation step and Covariance step terminated successfully;, ( ii ) 95% CI for Mixture partition did not include 0 nor 1; and, ( iii ) the change in the OBJ between mixture and non-mixture models was >5 . 99 ( χ2; p<0 . 05 , 2df ) ., In the present analysis , the difference was 19 . 8 ., The mixture model identified the interaction between azithromycin and ivermectin to be due to changes in bioavailability in Subpopulation B . Their mean estimate of bioavailability ( F ) was 1 . 37 relative to baseline , whereas F was unchanged for Subpopulation A ( 0 . 97 ) ., Inspection of noncompartmental data for Subpopulation B were consistent , showing higher Cmax and earlier Tmax values ( Cmax A: 54 . 3 ng . h/mL; B: 67 . 8 ng . h/mL; Tmax A: 4 . 1 h; B: 3 . 4, h ) ., There were no differences in apparent clearance or volume of distribution ., However the mechanism for the increase in bioavailability is unclear ., Azithromycin , like ivermectin , is a substrate for Pgp , however it has minimal inhibitory effects on this transporter in vitro 20 ., Although ivermectin is extensively metabolized by CYP3A4 16 , azithromycin has no inhibitory activity against this enzyme 22 ., There are no other plausible metabolic or transporter mechanisms that could explain an interaction , and no clinical covariates were identified that characterized either subpopulation ., In addition , mean pharmacokinetic parameters of ivermectin were similar in both subpopulations in the baseline phase ( mean AUC A: 1019; B: 805 ng . h/mL; Cmax A: 52; B: 45 ng/mL; Tmax A: 5 . 3; B: 4 . 8h ) ., The model was used to simulate the range of peak ivermectin concentrations that might be encountered if azithromycin and ivermectin were co-administered ., These simulated data were then compared with the Cmax data reported in the high-dose ivermectin safety study 7 ., The median simulated Cmax data ( 46 . 0 , 34 . 1 and 40 . 3 ng/mL for non-mixture model , mixture models A and B respectively ) were approximately 5–7-fold lower than the 261 ng/mL value reported by Guzzo et al 7 ., Indeed , the most extreme individual simulated values ( 201 . 2 , 115 . 3 and 175 . 5 ng/mL for non-mixture model , mixture models A and B respectively ) were still lower than the mean value reported in the high-dose study 7 ., These data give a high level of confidence that peak exposures that are predicted to occur if ivermectin and azithromycin were co-administered would never exceed mean values seen under high dose conditions 7 , and which in this study were safe and well tolerated ., In the Amsden et al interaction study 5 , ivermectin was dosed with food ( a high-fat breakfast ) ., Food has been shown to increase the bioavailability of ivermectin over 2-fold 7 ., Because dosing of patients in Africa is unlikely to be with high fat meals , extreme peak ivermectin concentrations would be half of those reported in the simulation ., Interestingly , simulations from both the mixture model and the non-mixture model had generally similar predictions of ivermectin exposures ( average estimates and variability ) ., Both models confirmed that the maximum concentration achieved in the interaction phase would not exceed 201 ng/mL ( Figure 7 ) ., In spite of adding two parameters to the non-mixture model; the final parameter estimates for both models were very similar ( Tables 1 and 2 ) ., The inflation of variability and projections of extreme values for both sets of simulations is a consequence of using 1000 replicates , where the chances of sampling from the very extreme values of random error distributions are more probable ., However predicting extreme high values , even if they are very rare , is very useful from a safety perspective , and provide a “worst case” scenario of any extreme high exposures that might be encountered in a clinical setting/trial during co-administration ., There are several important caveats to this analysis ., The data collected from the drug interaction study was not intended for population analysis , and a larger data set would have been desirable ., The use of a mixture model could be criticized on the basis that random variations in the data could be ascribed post hoc to population differences ., Indeed , although the mixture model identified two populations on the basis of different effects on bioavailability , it is unclear mechanistically what this difference might be due to ., Finally , modeling and simulation can advise but cannot supplant clinical data ., The findings from this study should be confirmed in further clinical or pharmacokinetic studies ., In conclusion , this analysis demonstrates the utility of a population model approach to analyze drug interaction data ., The mechanism for the interaction was identified ( an increase in bioavailability in one subpopulation ) ., The model was also used to simulate multiple clinical trials , to identify the maximum exposures that might be observed during co-administration , and provides confidence that the peak ivermectin exposures would never exceed mean exposures that have previously been shown to be safe and well tolerated . | Introduction, Methods, Results, Discussion | A recent drug interaction study reported that when azithromycin was administered with the combination of ivermectin and albendazole , there were modest increases in ivermectin pharmacokinetic parameters ., Data from this study were reanalyzed to further explore this observation ., A compartmental model was developed and 1 , 000 interaction studies were simulated to explore extreme high ivermectin values that might occur ., A two-compartment pharmacokinetic model with first-order elimination and absorption was developed ., The chosen final model had 7 fixed-effect parameters and 8 random-effect parameters ., Because some of the modeling parameters and their variances were not distributed normally , a second mixture model was developed to further explore these data ., The mixture model had two additional fixed parameters and identified two populations , A ( 55% of subjects ) , where there was no change in bioavailability , and B ( 45% of subjects ) , where ivermectin bioavailability was increased 37% ., Simulations of the data using both models were similar , and showed that the highest ivermectin concentrations fell in the range of 115–201 ng/mL ., This is the first pharmacokinetic model of ivermectin ., It demonstrates the utility of two modeling approaches to explore drug interactions , especially where there may be population heterogeneity ., The mechanism for the interaction was identified ( an increase in bioavailability in one subpopulation ) ., Simulations show that the maximum ivermectin exposures that might be observed during co-administration with azithromycin are below those previously shown to be safe and well tolerated ., These analyses support further study of co-administration of azithromycin with the widely used agents ivermectin and albendazole , under field conditions in disease control programs . | This paper describes the use of a modeling and simulation approach to explore a reported pharmacokinetic interaction between two drugs ( ivermectin and azithromycin ) , which along with albendazole , are being developed for combination use in neglected tropical diseases ., This approach is complementary to more traditional pharmacokinetic and safety studies that need to be conducted to support combined use of different health interventions ., A mathematical model of ivermectin pharmacokinetics was created and used to simulate multiple trials , and the probability of certain outcomes ( very high peak blood ivermectin levels when given in combination ) was determined ., All simulated peak blood levels were within ranges known to be safe and well tolerated ., Additional field studies are needed to confirm these findings . | mathematics/statistics, pharmacology/drug interactions | null |
journal.pcbi.1005322 | 2,017 | Modelling Systemic Iron Regulation during Dietary Iron Overload and Acute Inflammation: Role of Hepcidin-Independent Mechanisms | Iron is an essential element for the organism ., It plays a critical role in oxygen transport , DNA synthesis , mitochondrial energy metabolism and as a cofactor of numerous enzymes 1 , 2 ., However , excess free iron catalyzes reactions that result in the formation of reactive oxygen species and oxidative stress ., Hence , iron homeostasis must be maintained within a narrow range to provide sufficient iron for cellular function while preventing the generation of oxidative stress 3 ., Systemic iron homeostasis is predominantly controlled by the interaction of the liver produced hormone hepcidin with its receptor , the iron transporter ferroportin ( Fpn ) , resulting in the degradation of Fpn 4–7 ., Fpn is the only known cellular iron exporter 8 , 9 ., It controls iron export from duodenal enterocytes that take up dietary iron , from iron-recycling macrophages , and from hepatocytes that store iron ., Iron release from cells through Fpn requires the ferroxidases ceruloplasmin and/or hephaestin 10–12 ., Hepcidin is produced in response to iron availability ( via the BMP6/SMAD signaling pathway ) , erythropoetic demand ( via erythroferrone ) , hypoxia and inflammatory mediators ( via JAK/STAT signaling ) 13–17 ., Binding of hepcidin triggers Fpn internalization , ubiquitination and subsequent lysosomal degradation and thus causes iron retention in iron exporting cell types 18 ., In addition , Fpn expression is regulated at the transcriptional level by hypoxia-inducible factor-2alpha ( HIF2α ) in response to hypoxia and iron deficiency 19 , 20 as well as by BACH1 and Nrf2 in response to excess heme or oxidative stress 21 ., At the translational level its expression is controlled by iron regulatory proteins ( IRP ) , which bind to an iron responsive element ( IRE ) located in its 5’UTR 22 , 23 ., Furthermore , Fpn expression is controlled by miRNAs 24 , 25 ., Under physiological conditions serum iron is bound to the transport glycoprotein transferrin ., Transferrin saturation is used as a measure for the serum iron level ., The fact that hepcidin and Fpn expression are tightly regulated by numerous control mechanisms assures that physiological concentrations of transferrin-bound iron are maintained ., Increased transferrin saturation increases hepcidin expression 26–28 , leading to enhanced Fpn degradation and reduced duodenal iron absorption and iron mobilization from storage iron ., In the case of hereditary hemochromatosis , where genetic perturbations cause hepcidin deficiency , chronic iron overload and organ damage will develop 29 , 30 ., On the other end of the spectrum , mutations in genes causing hepcidin overexpression will induce chronic iron deficiency and anemia 31 ., Similarly , in an inflammatory setting , Fpn expression is decreased by hepcidin-dependent and hepcidin-independent mechanisms ., While hepcidin expression is increased by inflammatory cytokines , such as IL6 , causing Fpn degradation , Fpn transcription is additionally reduced 32–35 ., Both control mechanisms reduce serum iron levels ( hypoferremia ) , acting as an effective innate immunity mechanism that restricts access of microbes to iron 36 ., However , if inflammation persists , the lack of iron results in a reduced iron supply for erythropoiesis , causing anemia of inflammation , which is frequently observed in hospitalized patients ., So far it is not clear how the different control mechanisms quantitatively impinge on Fpn to maintain cellular or systemic iron homeostasis ., Studies in both , hepcidin knockout mice and mice with an engineered point mutation in Fpn that render it resistant to hepcidin binding ( Fpn C326S mice ) clearly demonstrate that LPS treatment results in a similar reduction of serum iron level compared to wild type mice 37 , 38 ., This indicates that hepcidin is dispensable for the generation of inflammation-induced hypoferremia under some conditions ., To dissect the quantitative contributions of hepcidin dependent post-translational and transcription-mediated Fpn control mechanisms under inflammatory conditions and to identify network components that contribute to the establishment of organ-specific iron pools , we generated a multi-scale model describing systemic iron homeostasis at the organ level ., We extended our previously established model for hepcidin regulation via the BMP6/SMAD and the IL6/STAT3 signaling pathways 39 by now considering organ iron pools , organ-specific Fpn mRNA and protein synthesis/degradation rates and the impact of Fpn levels on the iron export from organs into blood ( see Fig 1 ) ., The model was fit to own experimental data obtained in mice maintained on different iron diets or exposed to inflammatory stimuli ( peritoneal LPS injection ) as well as to previously published data by Lopes et al . 40 , Daba et al . 41 , Deschemin et al . 38 and Lesbordes et al . 42 ., The model was validated by correctly predicting the responses of iron pools and iron-related proteins to a combined stimulus of dietary iron-loading and LPS treatment ., Finally , we applied the model to assess the individual contributions of Fpn regulatory mechanisms to the response to dietary iron perturbations and inflammation and to analyze mechanisms for the establishment of organ specific iron-pools ., Fig 2 gives an overview of the different steps of the study ., Iron homeostasis in the body is maintained by intricate regulatory mechanisms of iron uptake , release and flux between compartments ., Systems biological approaches are required to fully understand the dynamics of this complex network ., So far mathematical models only described subsystems of iron metabolism , including liver iron metabolism 43 , 44 , intestinal iron absorption 45 , iron release from macrophages 46 or storage of iron in ferritin 47 ., To describe iron metabolism on a systemic level , Lopes et al . derived a whole-body model , which integrated iron fluxes between blood and different organs 40 ., This model revealed how iron fluxes and distribution of iron pools between organs change upon alterations in dietary iron levels ., However , the model by Lopes et al . lacked the underlying regulatory mechanisms responsible for maintenance of iron homeostasis and therefore it only described flux changes phenomenologically ( by fitting each condition separately ) ., Motivated by this , we set out to develop a whole-body model of iron homeostasis , which explicitly describes intra- and extracellular regulatory loops that sense and modulate iron fluxes between different compartments ., As a major means of regulation , we focused on the systemic regulation of iron metabolism by the hepcidin-Fpn regulatory axis ( see Introduction ) ., Our model allowed us to simultaneously fit time-resolved data for multiple experimental perturbations , and to dissect the impact of individual regulatory loops on systemic iron homeostasis ., The structure of our model is outlined in Fig 1 ., The concentration of iron in the non-cellular part of the blood stream ( Fe serum ) is the central hub ., From the serum , iron can be imported into various organs ., This raises the total intracellular iron pool in each organ ( Fe liver , Fe spleen , Fe bone marrow , Fe red blood cells , Fe duodenum ) ., Besides these major iron-containing organs , a substantial amount of iron can be found in the remaining body ., We modeled this by including a lumped iron pool that sums up all remaining intracellular iron pools in the body ( Fe other organs ) ., In several reverse reactions , iron can also be exported from peripheral organs into the blood ., Additionally , there is uptake of dietary iron in the duodenum and loss of iron from the duodenum and from the compartment ‘other organs’ , which represent loss via the shedding of enterocytes and skin cells desquamation , respectively ., Moreover , red blood cells receive iron from plasma through the bone marrow compartment and deliver iron into the spleen , e . g . into splenic macrophages that recycle iron from aging red blood cells ., Additionally , iron uptake by spleen macrophages due to ineffective erythropoiesis is included 48 ., The iron flux model initially described by Lopes et al . corresponds to the organs and iron fluxes ( dark red arrows ) depicted in Fig 1 ., In this study , we additionally considered that the export of iron from peripheral organs is controlled by the iron exporter Fpn which is predominantly located at the plasma membrane of three cell types: duodenal enterocytes , macrophages and hepatocytes ., Fpn expression is described separately for each organ and is controlled by three regulatory mechanisms ( see Introduction ) :, ( i ) inflammatory cues ( e . g . LPS ) reduce the transcription of Fpn mRNA;, ( ii ) intracellular iron levels enhance the translation of Fpn mRNA into protein;, ( iii ) the turnover of Fpn protein is enhanced by the soluble polypeptide hepcidin ., Hepcidin expression is activated by the iron-sensing BMP6/SMAD pathway and by an inflammatory signaling cascade , which involves production of cytokines ( primarily IL6 49 ) and the subsequent phosphorylation of STAT3 transcription factor in hepatocytes ., Taken together , our model describes several auto-regulatory loops controlling iron homeostasis as well as multi-layered regulation by inflammatory signals , all of which converge on the modulation of Fpn expression levels ., To describe these phenomena , we derived a system of 20 coupled ordinary differential equations ( see S1 Text ) ., For example , the dynamics of the bone marrow iron pool derived from the model topology in Fig 1 is given by, d F e b m d t = F s e r → b m - F b m → R B C - F b m → s p l ,, with Fser → bm , Fbm → RBC and Fbm → spl quantifying the inflow from serum into bone marrow and the outflows from the bone marrow into the RBCs and spleen compartments , respectively ., The individual reaction rates were mainly described using mass action kinetics ., For instance , it was assumed that most iron fluxes are proportional to the iron concentration in the originating compartment , e . g ., F s e r → b m = v b m F e s e r , F b m → R B C = v R B C F e b m F b m → s p l = v s p l F e b m , ( 1 ), were the rate constants vbm , vRBC and vspl are model parameters ( determined by fitting to the data , see next section ) ., Iron export from peripheral organs into blood is additionally assumed to be proportional to the respective Fpn level , e . g ., F d u o → s e r = u d u o F e d u o F p n d u o , for the flow of iron from the duodenum into the serum ., Thereby , we denote by vorgan the rate constants of iron import into the organs and by uorgan the rate constants of iron export into the serum ., Fpn expression is described using a standard model of mRNA transcription and protein translation ., Details about the Fpn regulation by LPS , iron and hepcidin can be found in S1 Text ., Hepcidin induction by IL6 and BMP6 signaling pathways was described by a previously calibrated model 39 , which considers signal integration on the hepcidin promoter using thermodynamic state ensemble approach ( in the following referred to as ‘hepcidin promoter model’ ) ., The computational model in SBML format is available in the Supplement ( see S1 Model ) ., The model contained many unknown parameters ( e . g . rate constants of the iron flows between compartments vbm , vRBC , vspl in Eq 1 , see previous section ) , which had to be estimated by fitting the simulation results to experimental data ., To reduce the complexity of this fitting problem , we assumed in some cases that homologous reactions in different compartments proceed with the same kinetic rate constants ( see S1 Text ) ., For instance , the degradation rate of Fpn mRNA and the Michaelis-Menten constant corresponding to Fpn mRNA inhibition by LPS is considered to be equal in all organs ., Furthermore , the kinetic parameters of the hepcidin promoter model were fixed to the values that we had previously determined using systematic perturbations in the HuH7 cell culture system 39 ., Nevertheless , 48 kinetic parameters remained to be estimated from the data ., Additionally , 20 scaling parameters were fitted to match a model formulated in absolute concentration units to experimental data given in arbitrary units ., The model was calibrated based on time-resolved data from male C57BL/6-mice , which were either measured as a part of this study or taken from the literature ( in total 344 data points ) ., As summarized in Table 1 the calibration dataset comprised three experimental perturbations , which were either applied alone or in combination: These experiments were mostly performed in wild type animals , but data from hepcidin knockout mice were also included ., Model simulations were performed by analytically calculating a steady state before perturbing the system by a change in diet or in inflammatory status ., The different experimental perturbations were mimicked in the model by changing parameter values or the initial conditions for the solving of the ordinary differential equations that describe the temporal dynamics of iron pools and regulatory proteins ( see section ‘Model derivation and implementation’ ) ., For example , a change in dietary iron corresponds in the model to a change in the parameter value describing the influx on iron into the duodenum ., Injection of LPS corresponds to a change in the initial condition for the LPS concentration to a nonzero value ., Iron pool sizes of compartments are given as μg per animal , with experimental data from individual mice scaled to a standard mouse weight of 25 g at 10 weeks of age ., Details on the simulation of tracer iron distribution are given in S1 Text ., Fitting of the simulated trajectories to the experimental measurements was done using a multi-start local optimizer and by minimizing the χ2 metric which is a weighed difference between model and data ( S1 Text ) ., The robustness of the model predictions was assessed by simulating them for multiple parameter combinations with a similar goodness-of-fit ., We describe the dynamics of iron pools and regulatory proteins using a set of ordinary differential equations ( see previous section and S1 Text ) ., The kinetic parameters in this model were unknown and had to be estimated by fitting the model simulations to experimental data ., To generate the data necessary for model calibration and for the testing of the fitted model , we performed time-resolved experimental measurements in mice ., We subjected 9–10 weeks old male C57BL/6-mice to an intra-peritoneal LPS injection ( 1μg/g body weight ) and followed the dynamics of iron-related parameters ., At defined time points post injection ( 0 . 5/1/2/4/6/8/24/36/48/72 hours in a first experiment and 6/18/48 hours in a second experiment ) , we measured the levels of iron in serum , liver , spleen , duodenum and red blood cells ., Complementary , we also fitted data of previous studies that analyzed temporal dynamics of whole-body iron metabolism in response to dietary iron content changes or upon injection of LPS or iron tracer 38 , 40 , 41 ., LPS injection ( 1μg/g body weight ) led to a transient drop in serum iron levels which was accompanied by iron accumulation in liver and spleen , but not in duodenum or red blood cells ( see blue lines in Fig 3A , 3B , 3D , 3J and 3K and S1 Fig ) ., In order to further characterize iron re-distribution , we quantified iron transport regulators in peripheral organs using qPCR and western blotting ., The iron exporter Fpn was downregulated at the protein level in liver and spleen which likely explains iron accumulation in these compartments ( blue lines in Fig 3G and 3L ) ., Fpn downregulation is expected to be controlled by hepcidin , as we found inflammatory signaling pathways controlling hepcidin expression ( serum IL6 , liver pSTAT3 ) to be activated upon LPS injection ( see Fig 3H and S1 Fig ) ., Accordingly , we found liver hepcidin mRNA expression to be upregulated upon LPS injection ( Fig 3C and S1 Fig ) ., This most likely translates into increased levels of bioactive hepcidin in the circulation , as hepatic mRNA levels typically closely reflect the amount of released hepcidin peptides 50–52 ., In addition , we also observed a strong downregulation of Fpn mRNA expression in liver and spleen ( blue lines in Fig 3F and S1 Fig ) , as has previously been reported 33 ., Hence , two independent pathways exist for Fpn regulation , a hepcidin-dependent post-transcriptional mechanism and a transcriptional mechanism that may be independent of hepcidin ., We fitted our mathematical model to the time-resolved measurements summarized in Table 1 using the procedure described in the Model section and S1 Text ., The final model fits ( Fig 3 , blue lines and S1–S4 Figs ) show a good quantitative match to the data: When averaged over all 344 data points ( characterized by the means and standard deviation of 2–6 individual replicates ) , the best fits were about one standard deviation from the experimental data ( χ2/N = 1 . 08 ) ., Furthermore , the model qualitatively reproduced the following key features of iron metabolism and accurately described their dynamics: From this we concluded that our mechanistic model accurately describes iron homeostasis under various experimental conditions ., We next challenged the model by testing its accuracy in predicting conditions not used for model calibration ( data summarized in Table 2 ) ., First , we tested how our model would predict the LPS-response in iron-loaded animals , a condition where both STAT3 and SMAD1/5/8-signaling are altered ., This was done by changing the value of the parameter that describes dietary iron uptake ( Fefood , see S1 Text ) by the same fold-change as the experimental increase in dietary iron ., Our model predicted that the LPS-induced dynamics of iron-related parameters for mice maintained on an iron-rich diet should be comparable to those maintained on a regular diet , albeit starting from a higher set point ( red lines in Fig 3 ) ., Experimentally we maintained male C57BL/6-mice on an iron-rich diet ( 20000 ppm iron ) containing 100 times more iron than the normal diet for 4 weeks , and subsequently injected a single dose of LPS ( 1μg/g body weight ) ., The experimental data confirm the predictions of the model: the measured dynamics matched the model prediction for most variables ( serum , liver and duodenum iron , liver hepcidin , BMP6 mRNA , liver pStat and pSmad , liver Fpn mRNA and protein ) ., Deviations of the model fit from the experimental data were observed for diet-induced changes in the initial concentrations of spleen iron , RBC iron , spleen Fpn protein ( Fig 3D , 3K and 3G ) ., We concluded that our model reflects the main features of the response of iron metabolism towards LPS stimulations under iron overload conditions ., To further validate the model , we challenged it with perturbations inside the regulatory network and compared the results to previously published data ., Two experimental strategies have been performed to block the hepcidin negative feedback loop: First , disruption of Bmp6 signaling specifically in liver was achieved by using data derived from conditional SMAD4 knockout mice 53 ., Second , a knockin mouse was established ( Slc40a1C326S ) , which harbors a Fpn mutation that can no longer bind to hepcidin 37 ., We reproduced these new conditions in the model by setting the SMAD expression level to zero ( this has an influence on the hepcidin expression level calculated via the hepcidin promoter model , see Model section ) or simultaneously setting the parameter values corresponding to the hepcidin effect on ferroportin degradation to zero ( k 2 l i v e r , k 2 d u o , k 2 s p l e e n , see Eq . 7 in S1 Text ) , respectively ., In both cases , the model predictions , increased serum and liver iron pools and decreased spleen iron content ( see Fig 4A ) , were confirmed by the experimental data 37 , 53 ., Moreover , as in the experimental data , Fpn resistance to hepcidin resulted in elevated hepcidin expression , whereas loss of SMAD-signaling resulted in a marked drop of hepcidin expression ( Fig 4A ) ., We next explored the organ-specific role of Fpn regulation by hepcidin in more detail , by simulating the response of the system under the assumption that hepcidin-resistant Fpn is expressed in a tissue-specific manner ., To this end , the parameter values corresponding to the hepcidin effect on ferroportin degradation ( k 2 l i v e r , k 2 d u o , k 2 s p l e e n , see Eq . 7 in S1 Text ) were separately set to zero ., Only the elimination of hepcidin-mediated Fpn regulation in the duodenum had a systemic effect on iron levels by increasing iron levels in serum , spleen , liver and the total body ., By contrast , the simulation of hepcidin-resistant Fpn in liver or spleen resulted only in a local effect with a decreased iron pool in the respective organ and minimal changes in the other organs ( Fig 4B ) ., We conclude that hepcidin shows its strongest effect on steady state iron pools by regulating Fpn in the duodenum , where iron is taken up and lost ., By contrast , Fpn level in liver and spleen predominantly affects the fraction of iron that circulates through these organs ., Mouse models with tissue specific hepcidin-resistance have not been described so far ., However , tissue-specific deletion of the FPN gene has been studied in Fpnflox/flox mice carrying an intestine-restricted villin-Cre transgene that is inducible by tamoxifen 54 ., Upon tamoxifen-induced expression , the Cre recombinase cuts out the intestinal FPN gene , and this leads to severe deficiency in the blood , liver and spleen 54 ., This also indicates a systemic effect of duodenal Fpn levels on the body iron pools ., We further compared changes in the dynamics of iron-related parameters in response to chronic inflammation predicted by the model to qualitative knowledge from the literature ., Chronic inflammation can cause anemia of inflammation , which is characterized by decreased serum iron and hemoglobin levels 55 ., In this disease , inflammatory cytokines induce hepcidin expression , which results in reduced Fpn-mediated iron export , thus limiting iron availability in the blood stream for erythropoiesis in the bone marrow ., If the inflammation persists anemia will develop as a consequence 56 ., Chronic inflammation was mimicked in the model by assuming a constant source term in the equation that describes the time derivative of the LPS concentration ( see Model section and Eq . 3 in S1 Text ) ., Our model reflects the known iron-related alterations of chronic inflammation: during persistent inflammatory stimulation by LPS , serum iron levels decreased by about 85% within 2 days , while RBC iron decreased over a longer time scale until it finally stabilized at about 10% of the normal level after two months ( Fig 4C ) ., In agreement with data reported in 57 , we find that the liver iron content increases in response to chronic inflammation ( Fig 4C ) ., The above described model analyses , comprising both fitting and prediction , show that our mechanistic model is able to accurately describe a broad spectrum of perturbations on the quantitative level in many cases and on the qualitative level in most cases ., Hence , our model allows us to mechanistically dissect systemic iron homeostasis and to evaluate to which extent hepcidin or Fpn contribute to the establishment of hypoferremia or maintenance of organ iron pools ., Fpn is regulated on the transcript and protein level in response to inflammation 9 , with hepcidin-dependent regulation considered to be the major contributor to hypoferremia ., As our model encompasses both mechanisms of Fpn regulation ( cell-autonomous transcriptional inhibition and hepcidin-mediated regulation ) , it allowed us to analyze the relative contribution of each mechanism separately by simulating the LPS response when either the transcriptional or the post-translational LPS effect on Fpn protein levels was eliminated ., The initial condition of the simulations was the non-perturbed steady state of the system with both regulatory mechanisms included ., Starting from this initial state , the reaction to LPS injection was simulated while keeping either hepcidin levels or Fpn mRNA levels constant ( the time derivatives of these model variables were set to zero , see Eqs . 1 and 6 in S1 Text ) ., Interestingly , the lack of the LPS mediated hepcidin induction showed an almost normal drop in serum iron levels ( 75% of the complete model , see Fig 5A , red line ) ., By contrast , the removal of the Fpn transcriptional control in response to LPS had a stronger effect and alleviated hypoferremia to 50% of the control ( see Fig 5A , green line ) ., The strong effect of inflammation-mediated transcriptional regulation of Fpn became even more evident in animals with dietary iron overload ., Removal of hepcidin regulation in this case led to a near to normal level of inflammation-induced hypoferremia , while the serum iron drop induced by hepcidin regulation alone was reduced ( Fig 5A ) ., This shows that hypoferremia arises by a combination of hepcidin-dependent and independent mechanisms with the total effect on the serum iron level being less than additive ., Even though the relative contribution of transcriptional control of Fpn expression seems dominant over the hepcidin mediated control of Fpn in the high-iron setting , hepcidin does play a critical role ., The increased basal level of hepcidin in iron loaded animals reduces the Fpn protein half-life ( via the terms k2hep , see Eq . 7 in S1 Text and S6 Fig ) ., This in turn couples Fpn protein levels more tightly to Fpn mRNA levels ( compare red and blue lines in Fig 5B ) ., As a result , induction of hypoferremia by transcriptional inhibition of Fpn and the recovery of serum iron levels are faster in iron loading conditions , with and without inflammatory control of hepcidin ( Fig 5A ) ., From this , we conclude that the transcriptional control of Fpn expression is the major determinant of the degree of hypoferremia ., It amplifies the effect of hepcidin-mediated protein degradation in an acute inflammatory setting ., We next focused our analysis on iron distribution among body compartments under conditions when mice were maintained either on an iron-enriched diet or on a regular diet ., Our data reflect the role of the liver as the main iron storage organ of the body , which is well described in the literature 58 , 59 ., Fig 6A shows the measured and fitted distributions of iron between the different pools in mice fed a normal or iron-enriched diet , respectively ( same experiment/data as in Fig 3 ) ., The iron content in all compartments except for the red blood cells increased after 4 weeks of dietary iron overload ., The liver shows both the highest relative and absolute change of iron content: a 10-fold increase in liver iron levels corresponding to an additional ∼600μg iron per mouse ( arrows in Fig 6A ) ., The model fits these data quantitatively ., We experimentally checked whether liver iron accumulation correlates with a low expression of the iron exporter Fpn ., Unexpectedly , hepatic Fpn protein levels were increased in mice fed with an iron-enriched diet , even though the BMP6/Hepcidin pathway was activated as expected ( S5A Fig ) ., Increased Fpn protein levels can be explained only partially by transcriptional regulation , as we observed only a modest increase in FPN mRNA levels ( S5A Fig ) ., While the macrophage marker F4/80 was increased in livers of LPS treated animals , this was not observed for iron-loaded animals compared to animals on a normal diet ( S5B Fig ) , suggesting that increased hepatic Fpn protein levels in iron-loaded animals are not explained by the infiltration of macrophages in the liver that express higher levels of Fpn compared to hepatocytes ., Thus , our data may suggest that the IRE/IRP-mediated translational regulation of Fpn expression is more pronounced than hepcidin-mediated Fpn regulation in the case of sustained dietary iron overload ., These experimental findings indicate that hepatic iron accumulation cannot be explained by hepcidin controlled Fpn degradation ., As a matter of fact , the increase in Fpn levels should counteract iron accumulation ., Hence , liver specific iron uptake must be assumed to be considerably higher to make up for the increased iron export ., To investigate the cause for the increase in liver iron levels , we tested additional Fpn-independent regulatory mechanisms for the iron exchange between liver and serum and performed model selection based on their fitting performance ., Two mechanisms have been included in the final model variant ( these were also present in all simulations shown above ) ., Fig 6B shows for the full model and models lacking one or both of these mechanisms the fitting performance of liver iron accumulation upon dietary iron overload and in HAMP-KO mice ., We find that the full model is the only one that fits all liver iron data ., We conclude that the hepcidin-Fpn axis alone is not sufficient to explain liver iron accumulation during dietary iron overload ., Chronic iron overload causes irreversible organ damage as exemplified by disease conditions such as hereditary hemochromatosis 67 ., Our own data and published data 14 , 41 show that even pronounced increases in dietary iron levels translate into a comparably small accumulation of iron in organs ( iron homeostasis ) ., Specifically , we observed that a 100-fold increase in dietary iron content results in a less than two-fold increase in most body iron pools ., The most dramatic change with a ∼10-fold increase occurs in the liver ( see arrows in Fig 6A ) ., Thus , all pools respond to dietary changes in a ( much ) less than proportional manner , both in vivo and in the model ( see Fig 6A ) ., To better understand the mechanisms that maintain iron homeostasis , we systematically perturbed each parameter of the model ( e . g . protein synthesis/degradation rates , iron import/export rates ) and analyzed which one shows the most pronounced impact on serum iron levels and induces the measured response of body iron pools to changes in the diet ( see S7 Fig ) ., This analysis reveals that limited iron uptake in the duodenum ( see below ) is the most critical mechanism to buffer all body iron pools against increases in dietary iron ., Apical uptake of iron from the lumen into duodenal enterocytes occurs via the divalent metal iron transporter 1 ( DMT-1 ) ., This process is regulated locally by cellular iron levels and hypoxia 68 ., Under iron rich conditions , iron absorption into enterocytes decreases due to downregulation of iron transporters by the IRE/IRP system and hypoxia-inducible factor 2 59 , 69–71 ., Based on this , we assumed in the model that iron uptake from the diet into duodenal enterocytes shows saturation ( this mechanism was also present in all simulations shown above ) ., This was implemented by using a Michaelis-Menten type equation to describe the duodenal iron uptake rate as a function of the dietary iron content ( see Eq . 14 in S1 Text ) ., The Michaelis-Menten constant of this equation ( parameter Kduo , see S1 Text ) was fitted to approximately five times the normal iron diet content which results in strong saturation for the 100-times iron enriched diet ., Eliminating this saturation by assuming a linear dependence of the absorption rate on the dietary iron content led to a loss of homeostasis for increasing dietary iron content ( compare blue and red lines in Fig 7 ) ., The model predicted that a 100-fold increase in dietary iron should result in a 2 . 5-fold increase in dietary iron uptake ( 40 and 100 μg iron per day in mice under normal and iron overload conditions , respectively ) ., We have indirectly tested this prediction by determining the iron contents in individual organs and the remaining carcass and summing up the individual pools for mice on normal and high iron diets ( Fig 6A ) ., In line with the hypothesis of iron-uptake saturation , | Introduction, Results, Discussion, Materials and Methods | Systemic iron levels must be maintained in physiological concentrations to prevent diseases associated with iron deficiency or iron overload ., A key role in this process plays ferroportin , the only known mammalian transmembrane iron exporter , which releases iron from duodenal enterocytes , hepatocytes , or iron-recycling macrophages into the blood stream ., Ferroportin expression is tightly controlled by transcriptional and post-transcriptional mechanisms in response to hypoxia , iron deficiency , heme iron and inflammatory cues by cell-autonomous and systemic mechanisms ., At the systemic level , the iron-regulatory hormone hepcidin is released from the liver in response to these cues , binds to ferroportin and triggers its degradation ., The relative importance of individual ferroportin control mechanisms and their interplay at the systemic level is incompletely understood ., Here , we built a mathematical model of systemic iron regulation ., It incorporates the dynamics of organ iron pools as well as regulation by the hepcidin/ferroportin system ., We calibrated and validated the model with time-resolved measurements of iron responses in mice challenged with dietary iron overload and/or inflammation ., The model demonstrates that inflammation mainly reduces the amount of iron in the blood stream by reducing intracellular ferroportin transcription , and not by hepcidin-dependent ferroportin protein destabilization ., In contrast , ferroportin regulation by hepcidin is the predominant mechanism of iron homeostasis in response to changing iron diets for a big range of dietary iron contents ., The model further reveals that additional homeostasis mechanisms must be taken into account at very high dietary iron levels , including the saturation of intestinal uptake of nutritional iron and the uptake of circulating , non-transferrin-bound iron , into liver ., Taken together , our model quantitatively describes systemic iron metabolism and generated experimentally testable predictions for additional ferroportin-independent homeostasis mechanisms . | The importance of iron in many physiological processes relies on its ability to participate in reduction-oxidation reactions ., This property also leads to potential toxicity if concentrations of free iron are not properly managed by cells and tissues ., Multicellular organisms therefore evolved intricate regulatory mechanisms to control systemic iron levels ., A central regulatory mechanism is the binding of the hormone hepcidin to the iron exporter ferroportin , which controls the major fluxes of iron into blood plasma ., Here , we present a mathematical model that is fitted and validated against experimental data to simulate the iron content in different organs following dietary changes and/or inflammatory states , or genetic perturbation of the hepcidin/ferroportin regulatory system ., We find that hepcidin mediated ferroportin control is essential , but not sufficient to quantitatively explain several of our experimental findings ., Thus , further regulatory mechanisms had to be included in the model to reproduce reduced serum iron levels in response to inflammation , the preferential accumulation of iron in the liver in the case of iron overload , or the maintenance of physiological serum iron concentrations if dietary iron levels are very high ., We conclude that hepcidin-independent mechanisms play an important role in maintaining systemic iron homeostasis . | medicine and health sciences, liver, immune physiology, pathology and laboratory medicine, spleen, immunology, messenger rna, diet, simulation and modeling, physiological processes, nutrition, signs and symptoms, homeostasis, duodenum, digestive system, research and analysis methods, inflammation, immune response, gastrointestinal tract, biochemistry, rna, diagnostic medicine, anatomy, nucleic acids, physiology, biology and life sciences | null |
journal.pbio.1002024 | 2,014 | A Structural Model of the Genome Packaging Process in a Membrane-Containing Double Stranded DNA Virus | The functional and structural knowledge of assembly principles of macromolecular complexes , in general , and viruses , in particular , have extended our understanding of viral capsid maturation and genome packaging processes ., The model systems used are most often double-stranded DNA ( dsDNA ) viruses composed of only proteins and nucleic acids ., Viruses with lipids possess additional complexity when exploring the mechanistic and structural properties of such fundamental functions ., The common mechanism for the genome encapsidation in icosahedral dsDNA viruses , including head-tailed phages , herpes , pox , and adenoviruses , involves a translocation of the viral DNA into a preformed procapsid by an ATP-driven reaction powered by the packaging complex localized at a single vertex 1 ., This single vertex-portal complex operates in both genome delivery and packaging ., A dodecameric connector at a 5-fold vertex provides a conduit for nucleic acid entry into the capsid 2–5 ., It is also an assembly site for the transiently associated packaging NTPase powering DNA translocation 6 ., The DNA packaging complex in tailless icosahedral dsDNA viruses with an internal membrane , such as bacteriophage PRD1 , operates in a similar manner , but is driven by a virion associated ATPase 7 , 8 ., In PRD1 , the ATPase P9 powers DNA packaging and has , in addition to the Walker A and B motifs , a conserved motif that may contribute to its anchoring to the membrane 7 ., P9 also shares sequence similarity with several other putative viral packaging ATPases , implying that this packaging mechanism might be common among the internal membrane-containing viruses 7 ., The only structural evidence for the packaging components of a tailless icosahedral virus with a membrane comes from the crystal structure of the archaeal Sulfolobus icosahedral virus 2 ( STIV2 ) packaging ATPase , which shows that these ATPases belong to the FtsK-HerA superfamily of P-loop ATPases , having both cellular and viral members 9 , 10 ., However , how the packaging complex is connected to the virion and how it provides a conduit through the internal membrane remain unknown ., The discovery that bacterial virus PRD1 and human adenovirus have the same major capsid protein ( MCP ) fold and virion architecture led to the hypothesis that viruses infecting host cells belonging to different domains of life are related , even though they do not share any detectable sequence similarity 11 , 12 ., This finding has led to the structure-based classification of viruses , and accordingly it was also proposed that viruses fall into a relatively small number of structure based viral lineages 13 , 14 ., One of these lineages is represented by PRD1 and includes several other viruses such as adenovirus , bacteriophage PM2 , vaccinia virus , Paramecium bursaria chlorella virus 1 ( PBCV-1 ) , archaeal Sulfolobus turreted icosahedral virus ( STIV ) , and virophage Sputnik 15–21 ., In addition , there are also similar viruses with two MCPs instead of one ., The relation of these viruses to the double β-barrel MCP containing viruses has been recently discussed 22 , 23 ., All these viruses are thought to derive from a common ancestor preceding the separation of the three domains of cellular life 13 , 24 , 25 ., Bacteriophage PRD1 is the best-studied viral system , where the virion possesses an internal membrane ( Figure S1 ) ., The broad structural information on PRD1 , down to atomic resolution , has provided insights into assembly principles of complex viruses 26–28 ., The mature virion ( ∼66 MDa ) is formed of at least 18 protein species of which ∼ten are membrane associated , constituting about half of the membrane mass 29 , 30 ., The external pseudo-T\u200a=\u200a25 icosahedral capsid shell of PRD1 is composed of 720 copies of the MCP P3 ( 43 . 1 kDa ) cemented together by 60 copies of minor coat protein P30 ( 9 . 1 kDa ) ( Figure S1A ) 26 , 31 ., The MCP P3 has a canonical double jellyroll fold , which is conserved within the lineage of PRD1-like viruses 11 , 26 ., The viral membrane , which is selectively acquired from the host plasma membrane , has a higher phosphatidylglycerol/phosphatidylethanolamine ( PG/PE ) ratio than that of its host 32 , 33 ., In addition , the lipids in the viral membrane are asymmetrically distributed between the leaflets—PE and PG are enriched in the inner and outer leaflets , respectively , most probably due to the high membrane curvature imposed by the capsid 27 , 33 ., In PRD1 , the regular 5-fold vertex ( receptor binding vertex ) consists of the membrane anchor protein P16 ( 12 . 6 kDa ) , penton base protein P31 ( 13 . 7 kDa ) , receptor recognition protein P2 ( 63 . 7 kDa ) , and spike protein P5 ( 34 . 2 kDa ) 26 , 31 , 34–37 ., Protein P2 initiates infection by attaching to the host cell receptor 35 , 38 ., However , unlike head-tailed bacteriophages in which the tail hub is used to penetrate the host cell envelope and provide a channel for genome delivery , PRD1 uses its internal membrane that transforms into a tail tube penetrating the capsid through an opening at the unique vertex and crossing the host cell envelope 38–40 ., The structural transition of the membrane triggers the release of the other vertex complexes leading to the loss of interaction between the capsid and the underlying membrane and allowing the tube to be formed 39 ., Among the 12 icosahedral vertices , PRD1 has one unique vertex responsible for the packaging of its linear 14 , 297 bp-long dsDNA genome , where the covalently 5′ end linked terminal proteins are necessary for genome packaging as well as for replication ( dsDNA-P8 complex; P8 is a 29 . 6 kDa protein ) 8 , 41 , 42 ., The unique vertex consists of transmembrane proteins P20 ( 4 . 7 kDa ) and P22 ( 5 . 5 kDa ) as well as proteins P6 ( 17 . 6 kDa ) and P9 ( 25 . 8 kDa ) , which were identified by genetic analyses and immuno electron microscopy ( Figure S1B ) 7 , 43–45 ., Previous experiments have shown that there are naturally occurring empty procapsids that lack protein P9 and are incompetent to package the genome 8 ., The in vitro packaging system applied to different PRD1 packaging mutants showed that while P9 is the packaging ATPase , the packaging efficiency factor P6 participates in the process , most probably by having a role in the incorporation of P9 into the unique vertex 7 , 8 , 45 ., To date , the unique vertex still remains structurally elusive , mainly due to technical difficulties in identifying non-icosahedral features in a highly symmetrical virus particle for cryo-EM structural determination ., In this study , we report the structure of a viral packaging complex with a membrane conduit using cryo-EM reconstruction without icosahedral symmetry imposition at 12 Å resolution ., Using virus particles devoid of specific unique vertex protein species allowed us to define the structure of this DNA translocation conduit and propose an assembly pathway for this portal structure crossing both the protein shell and the underlying viral membrane layer ., Resolving non-icosahedrally organized features that are essential functional components in icosahedral viruses remains a challenge ., Using algorithms specific to handling icosahedral objects in the multi-path simulated annealing ( MPSA ) software package 46 , 47 , several non-icosahedrally symmetric features in icosahedral viruses have been revealed , such as the tail organization in cyanophage P-SSP7 47 , the portal in herpes simplex virus 1 B-capsid 4 , and the portal in enteric phage P22 procapsid 5 and mature virion 2 , 48 ., In order to reveal the unique vertex in tailless mature PRD1 virion , 26 , 000 out of 50 , 000 particles were used to reconstruct the final density map at 12 Å resolution based on gold-standard criterion of two independent datasets 49 , 50 without icosahedral symmetry imposition ( Figures 1A–1C and S2A; Tables 1 , 2 , and S2; Movie S1 ) ., The map showed a unique packaging complex structure at one of its 12 vertices ( Figure 1C and 1D ) and regular 5-fold structures in the remaining 11 vertices ( Figure 1E ) ., The receptor recognition protein P2 and spike protein P5 were not resolved at the regular 5-fold vertices because of their flexible nature 37 ., Except for the unique vertex , the overall virion density map revealed a similar capsid organization as in the X-ray structure of the icosahedral PRD1 capsid 26 ., The Fourier shell correlation ( FSC ) calculated between the crystal structure of the MCP P3 ( PDB: 1W8X , chain B ) and the virion cryo-EM density map indicated that their structures match to 12 Å based on 0 . 5 FSC criterion ( Figure S2B ) ., This quantitative measure is substantiated by their apparent structural match ( Figure 1F ) and validates the overall accuracy of the image processing protocol ., Crystal structures of the penton protein P31 and the MCP P3 26 were docked into the cryo-EM density map ( Figures 2A and 2B; Movie S2 ) ., The docking shows unambiguously that the unique vertex does not have the pentameric protein P31 and the five neighboring MCP P3 trimers ( peripentonal MCPs ) as do the regular 5-fold vertices ( Figure 2A ) ., At unique vertex position , the packaging complex is surrounded by ten MCP P3 trimers ( Figure 2B ) ., The segmented unique packaging vertex comprises not only the capsid region that replaces the regular 5-fold structure , but also the transmembrane region that anchors the inner membrane layer interior to the capsid shell ( Figure 2C ) ., To understand the interactions between the unique packaging vertex and the capsid shell , the electrostatic inner surface of the ten P3 trimers surrounding the packaging complex was calculated by APBS 51 ., The inner surface of the surrounding MCPs had an overall weak negative charge , leading to a hypothesis that the outer surface of the packaging complex is positively charged to allow a stable interaction with the encompassing capsid shell ( Figure 2D ) ., As we have determined the overall structure of the unique packaging complex in the mature virion , the locations of the four packaging protein candidates remain unassigned in the complex ., We thus investigated the structures of the procapsid and three other packaging deficient mutant particles in order to localize the four protein species forming the packaging vertex ., Comparison of the mature virion to the procapsid devoid of packaging ATPase P9 and the viral genome ( dsDNA-P8 complex ) allowed the initial dissection of different protein components of the packaging vertex ., The procapsid density map without icosahedral symmetry imposition at 14 Å gold-standard resolution ( Figures 3A–3C and S3A; Tables 1 , 2 , and S2; Movie S3 ) revealed that the organization of the MCP and internal lipid membrane was similar to that of the icosahedral map of the procapsid 28 ., We noted a sharper fall-off of the FSC plot at low resolution between the two independent maps of the procapsid ( Figure S3A ) relative to that observed in the mature virion ( Figure S2A ) , which can be attributable to disordering of the lipid membrane in the procapsid ( Table 2 ) ., Docking of the crystal structure of MCP P3 into the symmetry-free procapsid density map ( Figure S3B ) revealed that their structures match ., The FSC between the P3 crystal structure and the segmented P3 cryo-EM density shows a structural match to 14 Å based on the 0 . 5 FSC criterion ( Figure S3C ) ., On the basis of the difference map calculated between the procapsid and the mature virion at equivalent resolution ( Figure S4A and S4B ) , the unique vertex of the procapsid displayed densities only in the transmembrane conduit but not at the radii of the capsid shell exterior to the membrane ( Figure 3B and 3C ) ., No density was observed on either side of the lipid bilayer confirming that protein P9 is part of the unique vertex ., Since P9 is considered to reside at the external surface of the virus 8 , we could attribute the missing density facing the exterior part of the virus to P9 ( Figure 2C ) ., To localize the packaging efficiency factor P6 in the unique vertex , we utilized packaging deficient mutant Sus621 particles ( amber mutation in gene VI ) , which are devoid of P6 and in which the amount of P9 is reduced to less than half of the wild-type ( wt ) amount ( Table 1 ) ., The density map of the Sus621 particle at 19 Å gold-standard resolution ( Figures 3D–3F and S5A; Table S2 ) revealed the transmembrane densities at the unique vertex similar to those seen in the procapsid ( Figure 3B and 3C ) ., The maps of the unique vertices in the procapsid and Sus621 particle lacked any density exterior to the membrane ( Figure 3C and 3F ) ., The icosahedrally arranged capsid proteins in the procapsid and Sus621 maps were structurally similar ., However , the regular vertex penton densities showed higher structural variance in the Sus621 mutant particle than in the procapsid , as shown in their difference maps both compared against the mature virion map ( Figure S4A–S4D ) ., These suggest that the regular vertex pentons in the mutant particle are not as rigid as that of the mature and procapsid particles yielding higher variance in the reconstructed densities ., When examining closely at the transmembrane densities at the unique vertices of the procapsid and Sus621 maps ( Figure 3C and 3F ) and their difference map at the same resolution ( Figure S6A ) , we found that there were extra densities in the center of the transmembrane densities in the procapsid map but not in the Sus621 map ., Since protein P6 is present in the procapsid but not in the Sus621 particle , these additional densities may correspond to the region of the P6 anchored to the center of the transmembrane conduit , while the remaining region of the P6 exterior to the membrane is disordered in the absence of P9 ., Hydrophobicity cluster analysis of the P6 sequence reinforces the presence of hydrophobic domains within protein P6 ( Figure S7A and S7B ) 52 , 53 ., To explain these observations , we propose that the density exterior to the membrane at the unique vertex is a composite of P9 and portion of P6 ., The non-membrane region of protein P6 is disordered in the procapsid lacking P9 , and protein P9 is disordered in the Sus621 particle in the absence of P6 ., When and only when P9 and P6 are both present , such as the case in the mature virion , they become well-ordered and their corresponding densities can be resolved ( Figure 2C ) ., Furthermore , the rest of the membrane density in the Sus621 particle ( Figure 3F ) appears to be less pronounced than that of the procapsid and the mature virion ( Table 2 ) ., This suggests that P6 may exert an impact on the membrane structure rigidity ., The low resolution fall-off in the FSC curve of the two independent maps in the Sus621 mutant particle ( Figure S5A ) also supports this interpretation of the membrane disordering ., In order to translocate the genome across the internal membrane of the virus , a transmembrane conduit has been proposed to exist at the unique vertex providing the channel for genome translocation 39 ., Secondary structure element predictions by psipred 53 indicate that proteins P20 ( 4 . 7 kDa ) and P22 ( 5 . 4 kDa ) both have one long transmembrane helix and one short one ( Figure S7C and S7D ) , implying that they can potentially form a transmembrane conduit at the unique vertex ., To assign protein components to the transmembrane region of the unique vertex , two packaging deficient PRD1 mutants ( amber mutation in gene XX or XXII ) were exploited ( Table 1 ) ., They are defective in the synthesis of protein P20 or P22 , and form unpackaged particles also lacking proteins P6 and P9 ( Sus526 and Sus42 particles ) ( Table 1 ) ., Based on biochemical analyses , it is not clear whether P20 and P22 are both simultaneously absent in the mutant particles 44 , 54 ., In the cryo-EM images of Sus526 and Sus42 particles ( Figure 4A and 4D ) , the membrane showed increased disorder and was unable to maintain a rigid shape ., With a new non-icosahedral symmetry particle orientation search approach ( details in Methods ) , we obtained the reconstructed density maps of the mutant particles determined at 22 Å and 18 Å gold-standard resolutions ( Figure S5B and S5C; Table S2 ) ., This revealed a void density in the capsid region and a disordered density in the membrane region at the unique vertex ( Figure 4B , 4C , 4E , and 4F ) , confirming that the unique vertex consists of proteins P6 , P9 , P20 , and/or P22 ., Based on the difference map calculated at the same resolution between the Sus526 particle and the procapsid ( Figure S6C and S6D ) and that between the Sus42 particle and the procapsid ( Figure S6E and S6F ) , the disordered transmembrane densities at the unique vertices of Sus526 and Sus42 particles do not contain the transmembrane conduit seen in the procapsid ., In addition , the rest of the membrane density in Sus526 and Sus42 particles appears to be weaker than that of the mature virion and procapsid ( Table 2 ) ., These observations suggest that proteins P20 and P22 contribute to the membrane density at the unique vertex and are critical to maintaining the integrity of the membrane ., Without the presence of P20 and P22 , the membrane exhibits additional flexibility ., Following the localization of the four protein species in the unique vertex , we examined the detailed features of the segmented packaging complex from the mature virion ( Figure 5A ) and the transmembrane conduit from the procapsid ( Figure 5D ) ., Exterior to the membrane region at the unique vertex in the mature PRD1 , the density is a composite of P6 and P9 and shows an apparent 12-fold symmetry based on rotational correlation curve ( Figures 5C and S8A ) ., It is surrounded by 10 MCP P3 trimers ( Figure 2B and 2C ) ., On the basis of the secondary structure prediction of P9 by psipred ( Figure S7E ) 53 , the packaging ATPase P9 has a conserved α/β phage portal motif 55 , suggesting that it can form a channel for genome translocation ., The density in the membrane region of the procapsid map displays an apparent 6-fold or 12-fold symmetry arrangement ( Figure 5F ) ., The volume of the transmembrane densities ( excluding the central extra density that could belong to part of protein P6 ) is estimated to be around 83 . 6 nm3 , which is equivalent to a molecular mass ∼69 kDa based on the previously established volume to mass equation 56 ., Six copies of P20 and six copies of P22 add up to 60 . 6 kDa , which is reasonably close to the observed density ., Based on rotational correlation curve ( Figures 5F and S8B ) , the peaks at 6-fold symmetry arrangement were higher than at 12-fold one , also suggesting that the density is organized as a hexamer ., Each of these hexameric components , potentially decorated by surrounding lipids , may represent a heterodimer made of one copy of P20 and P22 ., The central genome delivery channel , formed by P20 and P22 , is estimated to be 40–50 Å wide ., The assembly of P20/P22 complex may also provide the nucleating site for the packaging vertex assembly ., Interior to the membrane region , the packaging complex in the mature virion has an additional density , part of which probably corresponds to the terminal protein P8 complex with the dsDNA because this density is seen only in the virion ( Figure 5G ) ., A more close-up comparison of the density in the membrane region between the mature and procapsid maps showed some differences ( Figure 5H ) , which may be caused by the membrane bilayer itself undergoing a conformational expansion between the states of procapsid and mature virion 28 , 57 ., The structural change of the membrane may as well be induced by the addition of P9 , P6 , and DNA-P8 onto the packaging complex ., Many biological processes involve the utilization of ATP as the fuel source ., One exemplary illustration of the extensive roles of ATPases is the encapsidation of viral genomic material into a preformed procapsid shell ., PRD1 ATPase P9 provides the energy for the viral genome packaging as shown using an in vitro packaging assay 7 , 8 ., P9 has a dual role ., Functionally , it is a powerhouse to fuel the packaging process by hydrolyzing ATP , and structurally , P9 with the packaging efficiency factor P6 form the portal providing the external part of the channel at the unique vertex for DNA translocation ., The internal part of the packaging complex ( P20/P22 ) at the unique vertex is embedded in the membrane , and provides the transmembrane conduit ., P20/P22 complex also serves as the nucleating site for the whole specific vertex assembly ., The MCP P3 of PRD1 forms an icosahedral shell with a pseudo-T\u200a=\u200a25 lattice 11 , 26 ., At the regular 5-fold vertex , five P31 proteins organized as a penton with a strict 5-fold symmetry ( Figure 2A ) ., In the reconstruction without icosahedral symmetry imposition ( Figures 1 and 2 ) , there are 705 ( 720−5×3 ) P3 and 55 ( 60−5 ) P31 molecules forming the PRD1 capsid shell ., The special vertex occupied by several protein components does not obey 5-fold symmetry ( Figure 2B ) ., Ten MCP P3 trimers wrap around the 12-fold symmetrical P9/P6 complex at the unique vertex ( Figures 2B and 5A–5C ) ., Such a symmetry mismatch is a structural hallmark of the head-tailed dsDNA viruses with a portal complex arranged with 12-fold symmetry 3 , 4 , 47 , 48 ., In PRD1 mature virion , the internal membrane follows the shape of the icosahedral capsid shell ., This is presumably due to the pressure formed by the packaged genome , the presence of various membrane proteins , and the intercalation of the P3 shell into lipid moieties ( Figure S1B ) ., However , at the unique vertex , the proteins connected to the P9/P6 complex are membrane proteins P20 and P22 , which are organized with 6-fold symmetry ( Figure 5F ) ., A 12-fold versus 6-fold symmetry mismatch among different components at the unique portal vertex , seen here at the interface between P20/P22 and P9/P6 , is also found in membrane-less head-tailed dsDNA phages 47 ., We propose a molecular model for procapsid assembly and genome packaging ( Figure 6 ) , which will serve beyond PRD1 and provide one of the first structural clues in understanding the life cycle of the tailless internal membrane-containing icosahedral dsDNA viruses ., As the first step , newly-synthesized viral membrane proteins are transported to the cytoplasmic membrane of the host cell ( Figure 6A ) 58 ., The virus-specific membrane patch is then presumably pinched off , resembling the mechanism of the eukaryotic clathrin-coated pits , providing the framework for procapsid assembly ( Figure 6B ) ., The correct folding of certain viral structural proteins ( e . g . , MCP P3 ) and the formation of the PRD1 procapsid is facilitated by the host GroEL-GroES chaperonin and virus-encoded scaffolding protein P10 and assembly factor P17 ( and most probably P33 ) 59–61 ., Interestingly , procapsids devoid of the unique vertex can still assemble , which suggests that the membrane and/or other membrane proteins , for example , the membrane associated non-structural protein P10 31 , are functioning as a scaffold for capsid formation without the packaging complex ., However , lacking the P20/P22 membrane pore in the unique vertex leads to disordered internal membrane layers as suggested by the weaker intensities in the membrane region of the map ( Figure 4; Table 2 ) ., Since P20/P22 membrane pore is one of the defining features of the unique vertex , its absence leads to the formation of non-biologically active particles ., In these particles , the specific interactions between the capsid shell proteins and the membrane could be altered , which would result in a weaker density in the map ., In the procapsid , in which P9 is absent , and the Sus621 particle lacking P6 and half of P9 , the packaging process is deficient , but the presence of the P20/P22 conduit defines the unique vertex and thus allows the stable interactions between the capsid shell and the underlying membrane , making the internal membrane rigid ( Figure 3; Table 2 ) ., However , without the internal genome pressure , certain flexibility may exist in the membrane envelop of the procapsid and Sus621 particle ., Integral membrane proteins P20 and P22 , which tend to form hexameric heterodimers ( Figure 5F ) with potential lipid decorations , assemble to form a transmembrane conduit ( Figure 6C ) ., Then , P9 and the packaging efficiency factor P6 form a 12-fold portal complex with P6 positioned atop the transmembrane conduit ., P8 is linked to the 5′ end of the linear dsDNA genome and may recruit the genome to the packaging motor by binding to P9 ., After this complex is formed , the genome and genome-associated P8 begin to be packaged 45 ., Once the packaging efficiency factor P6 and packaging ATPase P9 together become ordered in their position in the unique vertex , DNA packaging can begin ., ATP hydrolysis by P9 provides the energy for DNA translocation into the procapsid through the unique vertex ( Figure 6D ) ., The conduit across the membrane formed by integral membrane proteins P20 and P22 provides the 40–50 Å wide channel for the dsDNA-P8 complex to be transported through the inner membrane underneath the capsid shell ., After packaging , the pore in the vertex must be sealed ., The terminal protein P8 may play a role as a protein valve similar to the valve of the head-tailed phage P-SSP7 47 ., After packaging the 14 . 9 kb dsDNA genome , the increased internal pressure leads to the expansion of the membrane , which dissipates the energy and prevents the massive expansion of the capsid shell ., The mature PRD1 virion , as observed in this study , has undergone membrane expansion ( Figures 1C and 3C; Table 2 ) ., The spacing of the lipid bilayer decreases upon the maturation of the particle and the membrane layer gets closer to the capsid shell as seen in our symmetry-free reconstructions as well as in the previous icosahedral maps ( Table 2 ) 57 , 62 ., In addition , the internal genome pressure and the closer interaction between the membrane and the capsid shell make the membrane envelope most secure in its relative position and thus result in a stronger density in the reconstructed map ( Table 2 ) ., These observed changes accommodate the packaging process and eventually lead to the maturation of PRD1 procapsid into infectious virion ( Figure 6E ) ., Biochemical and structural analyses of the unique vertices in the head-tailed dsDNA bacteriophages such as T4 , T7 , ϕ29 , P22 , epsilon15 , P-SSP7 , and some eukaryotic viruses have demonstrated that their packaging and assembly processes share similarity both functionally and structurally ., One example is the well-studied bacteriophage P22 2 , 5 , 48 , where the portal proteins function as the nucleating site for the procapsid assembly with the help of scaffolding proteins ., Once the procapsid is formed , the DNA is packaged through the channel of the portal powered by the terminase motor with ATPase activity 63 ., Virus maturation involves the release of the scaffolding proteins and terminase 64 , before the tail is attached at the unique vertex ., In phage ϕ29 65 , the MCPs , connector/portal protein , head fiber proteins , and packaging RNA ( pRNA ) molecules together form the prohead with the help of the scaffolding proteins ., Then the ATPase motor of ϕ29 packages the DNA into the prohead through the channel provided by the portal proteins and the pRNAs at the unique vertex ., After that the tail is attached onto the unique vertex completing the assembly of the virion ., The assembly of PRD1 differs from the head-tailed viruses ., First , PRD1 does not possess a conventional portal protein like the portal of phage P22 66 or the connector in ϕ29 67 ., How is the PRD1 procapsid formed ?, The portal protein complex of P9 and P6 are assembled to the procapsid , providing the channel for DNA translocation ., The ATPase activity of P9 provides the energy for DNA packaging 7 analogous to the terminase in P22 68 or the ATPase in ϕ29 69 ., However , PRD1 P9 does not dissociate from the capsid after the DNA is packaged as in P22 and ϕ29 systems ., Second , the packaging efficiency factor P6 of PRD1 serves as a facilitator for the ATPase motor in genome packaging ., In contrast , the DNA-packaging motor of bacteriophage ϕ29 is geared by a ring of pRNAs 70 ., Third , our study provides the structural insights into the packaging and assembly process in an icosahedral virus with an internal membrane ., In this membranous virus , P20 and P22 form a transmembrane nanotube and provide a nucleating site for the recruitment of P9 and P6 ., For comparison , in the head-tailed bacteriophages like P22 , the portal complex is the initiating site for procapsid assembly 5 ., Finally , during maturation of the head-tailed dsDNA bacteriophages , such as HK97 71 or P22 5 , the viral capsid goes through significant conformational changes including capsid expansion and angularization ., In contrast , virus maturation in PRD1 mainly involves the membrane expansion and conformational changes at the MCP-membrane interface as well as in the transmembrane densities at the unique vertex without major conformational changes in the viral capsid 57 , 62 ., Thus , it is the inner membrane in PRD1 that undergoes most of the significant structural re-arrangements during virus maturation , not the viral capsid shell ., The unique vertex of PRD1 resolved here portrays the detailed structural picture to advance our understanding on procapsid assembly and genome packaging in a membrane-containing virus ., The number of different PRD1-like icosahedral internal membrane-containing viruses is increasing: these can infect archaea , bacteria , and eukaryotes , covering all domains of life 72 ., This is the first time , to our knowledge , that such a packaging-portal complex structure is revealed ., Based on sequence data , all PRD1-like viruses encode a packaging ATPase , including the Walker A and B motifs and the P9-specific region 7 like PRD1 P9 ., However , even within this group of viruses the packaging mechanisms must differ between those with a circular or linear genome ., For viruses with a linear genome , the packaging mechanism resembles that of PRD1 , but for circular genomes , like in bacteriophage PM2 , the mechanism for the packaging/condensation of the genome could be totally different 16 ., Wt PRD1 and its packaging deficient mutants ( Table 1 ) were propagated ( LB medium at 37°C ) on Salmonella enterica serovar Typhimurium LT2 DS88 ( wt non-suppressor host ) 73 and on S . enterica suppressor strain PSA ( supE ) 74 or DB7154 ( supD10 ) 75 harboring plasmid pLM2 ., The suppressor-sensitive mutant phenotypes were verified by an in vivo complementation assay using plasmids carrying the corresponding PRD1 wt genes ( Tables 1 and S1 ) ., To reduce the background in mutant virus productions , the infected cells ( multiplicity of infection 8 ) were collected 15 minutes post infection ( Sorvall SLA3000 rotor , 5 , 000 rpm , 10 min , 22°C ) and resuspended in pre-warmed fresh medium ., Released virus particles were concentrated and purified by polyethylene glycol-NaCl precipitation , rate zonal ( 5%–20% gradient; Sorvall rotor AH629 , 24 , 000 rpm , 55 min , 15°C ) , and equilibrium ( 20%–70% gradient; Sorvall rotor AH629 , 24 , 000 rpm , 16 h , 15°C ) centrifugations in sucrose using 20 mM potassium phosphate ( pH 7 . 2 ) , 1 mM MgCl2 buffer 76 ., The equilibrated particles were concentrated by differential centrifugation ( Sorvall rotor T647 . 5 , 32 , 000 rpm , 2 h , 5°C ) and resuspended in the same buffer ., The protein concentrations were measured by Coomassie blue method using bovine serum albumin as a standard 77 ., The wt/revertant backgrounds of the purified mutant particles were analyzed by assaying their specific infectivity on suppressor and wt hosts ( Table 1 ) ., The prot | Introduction, Results, Discussion, Materials and Methods | Two crucial steps in the virus life cycle are genome encapsidation to form an infective virion and genome exit to infect the next host cell ., In most icosahedral double-stranded ( ds ) DNA viruses , the viral genome enters and exits the capsid through a unique vertex ., Internal membrane-containing viruses possess additional complexity as the genome must be translocated through the viral membrane bilayer ., Here , we report the structure of the genome packaging complex with a membrane conduit essential for viral genome encapsidation in the tailless icosahedral membrane-containing bacteriophage PRD1 ., We utilize single particle electron cryo-microscopy ( cryo-EM ) and symmetry-free image reconstruction to determine structures of PRD1 virion , procapsid , and packaging deficient mutant particles ., At the unique vertex of PRD1 , the packaging complex replaces the regular 5-fold structure and crosses the lipid bilayer ., These structures reveal that the packaging ATPase P9 and the packaging efficiency factor P6 form a dodecameric portal complex external to the membrane moiety , surrounded by ten major capsid protein P3 trimers ., The viral transmembrane density at the special vertex is assigned to be a hexamer of heterodimer of proteins P20 and P22 ., The hexamer functions as a membrane conduit for the DNA and as a nucleating site for the unique vertex assembly ., Our structures show a conformational alteration in the lipid membrane after the P9 and P6 are recruited to the virion ., The P8-genome complex is then packaged into the procapsid through the unique vertex while the genome terminal protein P8 functions as a valve that closes the channel once the genome is inside ., Comparing mature virion , procapsid , and mutant particle structures led us to propose an assembly pathway for the genome packaging apparatus in the PRD1 virion . | The life cycle of a virus involves serial coordination of viral molecular machines ., These machines facilitate functions such as membrane fusion and genome delivery during infection , and capsid formation and genome packaging during replication and shedding ., Icosahedral dsDNA viruses use one genome-translocation machine for both genome delivery and packaging ., The genome-translocation machine of the membrane-containing bacterial virus PRD1 is composed of four packaging protein species at a unique vertex ., Because these proteins do not follow the dominating icosahedral symmetry of the viral capsid , the structure of this vertex has remained elusive ., In this study , we localize the unique vertex in the virus from raw electron cryo-microscopy images of the virus ., We show that the genome-packaging complex of PRD1 replaces the regular 5-fold structure at the unique vertex and contains a transmembrane conduit as a genome translocation channel ., We extend our structural studies to the procapsid—a precursor of the virus—and three packaging mutant particles , allowing us to localize all individual protein species in the complex ., Based on these structures , we propose a model of the molecular mechanism of assembly and packaging in the life cycle of the PRD1 virus . | biochemistry, microscopy, proteins, virology, protein structure, electron microscopy, biology and life sciences, microbiology, computational biology, viral structure, research and analysis methods | Modeling how PRD1, a dsDNA membrane-containing virus, packages its genome using electron cryo-microscopy. |
journal.pcbi.1003597 | 2,014 | Modelling Negative Feedback Networks for Activating Transcription Factor 3 Predicts a Dominant Role for miRNAs in Immediate Early Gene Regulation | The concept of immediate early genes ( IEGs ) was initially established in relation to viral infection of bacterial or eukaryotic cells ( e . g . herpesvirus 1 ) ., Here , IEGs are defined as the genes encoding the first phase of mRNAs expressed from the viral genome , relying entirely on pre-existing host proteins ., Early and delayed genes are expressed later and require production of new proteins ., In mammalian cells , IEGs became defined as those which are regulated by pre-existing transcription factors ( TFs ) such that increases in IEG mRNA expression are not suppressed by protein synthesis inhibitors ( e . g . cycloheximide ) 2 ., Many IEGs encode transcriptional regulators that are required to modulate downstream gene expression ( i . e . second phase genes ) and some are inhibitory factors required to terminate transcription ., With the discovery of miRNAs that influence mRNA expression and translation 3 , 4 , some adjustment of the IEG concept is necessary according to whether mRNA regulation is transcriptional or post-transcriptional ., Synthesis , production and degradation of miRNAs may be insensitive to inhibitors of protein synthesis as would be any ( post-transcriptional ) effects on mRNA expression ., Cardiomyocytes , the contractile cells of the heart , withdraw from the cell cycle perinatally ., Neonatal rat cardiomyocytes are therefore highly synchronized and , because they do not divide , form a good model for the study of IEG regulation ., In these cells , endothelin-1 ( ET-1 ) , a Gq protein-coupled receptor ( GqPCR ) agonist , elicits maximal activation of the entire pool of extracellular signal-regulated kinases 1/2 ( ERK1/2 , the prototypic mitogen-activated protein kinases ) within 3–5 min 5 , 6 ., The earliest IEGs are substantially and maximally upregulated within 15–30 min , with second phase RNAs being detected within 1 h 7 ., ERK1/2 play a major role in the response and promote upregulation of ∼70% of ET-1-responsive transcripts 7 , 8 ., Recent data indicate that ERK1/2 act together with their downstream substrates p90 ribosomal S6 kinases ( RSKs ) to regulate RNA expression , with RSKs being required to increase expression of ∼50% of the RNAs upregulated by ET-1 9 ., As in other systems , upregulation of cardiomyocyte IEG mRNAs is acute and transient ., For example , activating transcription factor 3 ( Atf3 ) and early growth response 1 ( Egr1 ) mRNAs ( each regulated via ERK1/2 ) are both upregulated maximally by ET-1 within ∼30 min 10 ., However , expression of Egr1 mRNA returns to basal by ∼2 h and expression of Atf3 mRNA declines close to basal levels over ∼4 h ., This raises the question of negative feedback regulation ., Atf3 is emerging as an extremely important feedback regulator of transcription ., Particular emphasis is placed on its role in inflammation and Atf3 is essential for restraining the immune response , but Atf3 is also associated with cancer and cardiac dysfunction 11–14 ., However , although it is stress-regulated and stress-regulating , Atf3 is upregulated in many systems by growth stimuli including peptide growth factors and GqPCR agonists 7 , 15–17 ., Atf3 forms homo- or heterodimers that bind to ATF/CRE sites in gene promoters to regulate transcription 18 ., It is largely regulated at the level of expression , being present at very low levels in quiescent cells and induced as an IEG by a range of extracellular stimuli and cellular stresses ., ERK1/2 are particularly implicated in promoting Atf3 mRNA expression and several TFs may be involved 19–21 ., Atf3 is generally viewed as a transcriptional repressor , particularly when acting as homodimers , and it may repress transcription from its own promoter to limit expression 22 ., In cardiomyocytes , Atf3 operates in a negative feedback system with Egr1 and , by binding to the Egr1 promoter , Atf3 inhibits Egr1 transcription 10 ., Mathematical modelling of the system demonstrated that this , in itself , could suffice for the transient nature of the Egr1 response ., The pivotal role that Atf3 plays in biological systems renders it essential to understand the mechanisms that regulate Atf3 expression ., Here , we have developed our original mathematical model to determine whether Atf3 serves as a negative feedback regulator of its own transcription ( as suggested 22 ) and to explore other mechanisms that may switch off Atf3 mRNA expression ., We demonstrate that self-regulation by Atf3 on its own promoter ( or , indeed , self-regulation by any transcription factor ) is implausible and miRNAs are likely to play a dominant role in switching off Atf3 expression post-induction ., Previous studies used a protein overexpression approach with reporter assays for the Atf3 promoter to provide evidence that Atf3 operates in an autorepression loop and , by binding to an element immediately downstream of the TATA box , it inhibits its own transcription 22 ., Since it is difficult to test this experimentally with an endogenous system , we extended the mathematical model developed for the Atf3-Egr1 feedback system to determine whether a direct Atf3 autorepression loop is feasible ( Figure 2A ) ., The development of this and subsequent mathematical models is detailed in Text S1 ., The model extension requires three new parameters ., The association constant for Atf3 protein with Atf3 DNA ( K11 ) was assumed to have an initial value of 0 . 1 nM ., The forward and reverse rates of the Atf3 protein for Atf3 DNA , λ1 and λ−1 , were also required ., λ−1/λ1 was assumed to have a similar value to K11 , but a slow rate of reversal giving a value of λ1\u200a=\u200a1×105 ( Ms ) −1 ., With the initial parameters , there was little effect on Atf3 mRNA , Atf3 protein or Egr1 mRNA profiles ( Figure 2 , B–D , solid lines ) compared with the original model ( Figure 1K ) ., Decreasing the disassociation constant of Atf3 protein and Atf3 DNA by 106 produced profiles that were more representative of the experimental data for Atf3 mRNA and protein ( Figure 2 , B and C , dashed lines ) , but acute inhibition of Egr1 mRNA expression was lost ( Figure 2D , dashed line ) ., Decreasing the ratio of λ−1/λ1 ( informed by a sensitivity analysis ) 2 orders of magnitude had no effect on the results ., Increasing the ratio 2 orders of magnitude decreased the amount of Atf3 and increased the amount of Egr1 as expected ( not shown ) ., We conclude that Atf3 is unlikely to act in an autorepressive manner on its own promoter ., If it did , the rate of accumulation of Atf3 mRNA/protein would be reduced such that it could then not significantly influence other gene promoters ( in this case Egr1 ) producing a redundant system ., Thus , the Atf3 autorepressive loop is implausible ., Exposure of cardiomyocytes to BI-D1870 , an inhibitor of RSKs 31 , substantially enhances upregulation of Atf3 mRNA by ET-1 ( Figure 3A ) ., Although BI-D1870 alone increase ERK1/2 phosphorylation in cardiomyocytes , this over a similar time and to a lesser degree than that induced by ET-1 and it does not affect the activation of ERK1/2 by ET-1 9 ., It is therefore unlikely that the enhancement of Atf3 mRNA expression induced by ET-1 seen in the presence of BI-D1870 is due to its effects on phosphorylation of ERK1/2 and suggests that a negative feedback system downstream of RSKs ( or other kinases that may be inhibited by BI-D1870 ) moderates Atf3 mRNA expression ., Interestingly , the effects of BI-D1870 over 2 h ( Figure 3B ) produced a profile for Atf3 mRNA expression that resembled the predicted profile for Atf3 mRNA in the modifications to the original model ( Figure 1K ) suggesting that an input from RSKs might suffice to switch off Atf3 transcription ., We therefore extended the mathematical model to test the hypothesis that RSKs ( as exemplar kinases ) phosphorylate nuclear TFs ( RTF ) to negatively regulate Atf3 transcription ( Figure 3C ) ., Experimentally , maximal phosphorylation of RSKs occurs at 5–15 min , declining to ∼25% the maximal level by ∼30 min ( Figure 3D ) ; this profile was modelled mathematically ( Figure 3E , cyan line; see Text S1 for details ) ., The predicted profile for phospho-RTF binding to the Atf3 promoter is shown in Figure 3E ( red line ) ., Because different signals are applied to different proteins , we assumed that the negative signal from phospho-RSK and phospho-RTF was competitive with the positive signal from phospho-ERK1/2 and phospho-TF ., New parameters were required for the rate of RSK phosphorylation by phospho-ERK1/2 ( k9 ) , the rate of RSK-P dephosphorylation ( d10 ) , rates of association and disassociation of RTF for Atf3 DNA ( λ2 and λ−2 ) and the rates of phosphorylation and dephosphorylation of RTF bound to Atf3 DNA ( k10 and k−10 ) ., A fit-by-eye to the RSK data ( Figure 3D ) yielded k9\u200a=\u200a1×105 ( Ms ) −1 and d10\u200a=\u200a5 . 9×10−4 s−1 ., We assumed that λ2\u200a=\u200aλ1 and λ−2\u200a=\u200aλ−1 , k10\u200a=\u200ak5 and k−10\u200a=\u200ak−5 , and that the total amount of RSKs is equal to the amount of ERK1/2 ( i . e . R0\u200a=\u200aE0 ) ., With the initial parameters , the profiles for Atf3 mRNA and protein remained similar to the original model ( Figure 1K ) , although the amount of each was reduced ( Figure 3 , F and G ) and acute inhibition of Egr1 mRNA remained ( Figure 3H ) ., With the competition between phospho-TF and phospho-RTF for Atf3 DNA , increasing the rate of association of phospho-RTF with the Atf3 promoter may produce a profile that approaches that of the experimental data , but acute inhibition of Egr1 mRNA expression becomes lost ( not shown ) ., We conclude that the time delay between activation of ERK1/2 relative to RSKs is too small ( 1–2 min ) to result in differential effects on mRNA expression over 0 . 5–4 h if both act directly on transcription from the same promoter ., A greater time delay may be elicited by termination of transcription by an IEG encoding an inhibitory TF ( ITF ) and/or upregulation of a miRNA ( or another factor ) promoting Atf3 mRNA degradation and inhibiting Atf3 protein synthesis ., We next tested the hypothesis that downregulation of Atf3 mRNA expression ( we retained the assumption that this is via RSKs ) is mediated by upregulation of an ITF ( Figure 4A ) , making the assumption that the negative effect of ITF was competitive with the positive signal from ERKs ., Five further parameters are required for this model: the disassociation constant of ITF protein for Atf3 DNA ( K15 ) , the rate of ITF transcription ( k13 ) , the rate of translation of ITF protein ( k14 ) , ITF mRNA degradation ( d4 ) , and the rate of ITF protein degradation ( d5 ) ., On the basis that ITF is an IEG , we assumed that the disassociation constant of ITF protein was equal to that of TF for Atf3 DNA ( K15\u200a=\u200aK10 ) , rates of transcription and translation of ITF are similar to those for Atf3 ( i . e . k13\u200a=\u200ak6 and k14\u200a=\u200ak8 ) , and the rates of degradation of each mRNA and protein are similar ( i . e . d4\u200a=\u200ad2 and d5\u200a=\u200ad3 ) producing the profiles for ITF mRNA and protein shown in Figure 4B ., With initial conditions in which ITF binding to Atf3 DNA is similar to Atf3 binding to the Egr1 promoter ( Figure 4 , C and D , solid lines ) , there was little change in the profiles compared with the original model ( Figure 1K ) ., Increasing the rate of binding of ITF to Atf3 DNA by 106 produced profiles that were more similar to the experimental data ( Figure 4 , C and D , dashed lines ) , but inhibition of Egr1 transcription was reduced ( Figure 4E , dashed line ) ., Transcriptomics studies using Affymetrix exon arrays 9 identified several miRNAs that are regulated in cardiomyocytes by ET-1 within 1 h and whose upregulation is inhibited by BI-D1870 ( Table 1; since we have not performed miRNA profiling of cardiomyocytes , this is unlikely to be a full representation of miRNAs in this category ) ., Atf3 expression may be regulated by miRNAs ( e . g . Mir663 suppresses Atf3 mRNA expression in endothelial cells subjected to shear stress 32 ) ., We therefore tested the hypothesis that upregulation of one or more miRNAs ( we retained the assumption that this is downstream of RSKs as exemplar kinases ) is required to downregulate Atf3 mRNA expression ( Figure 5A ) ., This extension to the model introduces eight new parameters ., With little/no published information on rates of miRNA turnover , we informed the model as follows ., The rate of phosphorylation/dephosphorylation for RTF binding to miDNA was assumed to be equal to the rate of phosphorylation/dephosphorylation of TFs bound to Egr1 and Atf3 DNA by ERK-P ( k16\u200a=\u200ak3; k−16\u200a=\u200ak−3 ) ., The rate of miDNA transcription was assumed to be the same as that of Atf3 transcription ( k17\u200a=\u200ak6 ) and the rate at which premiRNA is converted to miRNA ( k18 ) was assumed to be double the rate of Atf3 translation ( 0 . 5/s ) ., The rate at which miRNA associates with Atf3 mRNA ( k19 ) was assumed to be equivalent to the rate set for Atf3 protein binding to Egr1 DNA ( 1×105 ( Ms ) −1 ) and the reverse rate ( k−19 ) was set at 5×10−5/s , the value that gave the best qualitative fit to the experimental data ., The rates of decay of premiRNA ( d6 ) and miRNA ( d7 ) were assumed to be ∼24 h in line with recently published data 33 , 34 ., The rate of decay of the complex miRNA·mRNAAtf3 ( d8 ) was initially assumed to be the same as that of Atf3 mRNA ( d8\u200a=\u200ad3 ) ., Mathematical modelling for the production of miRNA is shown in Figure 5B ., miRNAs may increase mRNA degradation ( within a complex with Ago2 ) or silence mRNA expression ( within Ago1 , Ago3 or Ago4 ) ., Computational modelling has been used to predict/confirm that the increase in mRNA degradation induced by miRNAs lies within the range 1 . 3- to 6 . 4-fold 35 ., We first assumed that potential miRNA ( s ) had a negligible effect on the rate of degradation ( i . e . the rate of degradation of the miRNA/mRNA complex was similar to the free mRNA ) ., In this case , the total Atf3 mRNA ( i . e . free Atf3 mRNA + miRNA/mRNA complex ) dynamical profile did not match the experimental data ( Figure 5C , solid line ) ., However , since the miRNA/mRNA complex is assumed to be translationally incompetent , the protein profile was significantly affected and approximated to the experimental profile ( Figure 5D , solid line ) ., If we assumed that potential miRNA ( s ) increased the rate of degradation of Atf3 mRNA from the miRNA/mRNA complex 2-fold , the total Atf3 mRNA profile approached that of the experimental data ( Figure 5C , long dashed line ) ., Increasing the rate of degradation 4-fold gave a profile that approximated to the experimental data ( Figure 5C , short dashed line ) ., The protein profile was relatively unaffected by increasing the rate of degradation of the miRNA/mRNA complex ( Figure 5D ) , indicating that miRNAs have a dominant effect on protein expression irrespective of their effect on mRNA expression ., In all cases , the predicted Egr1 mRNA profile approximated to the experimental data over the 2 h period studied ( Figure 5E ) , although subsequently Egr1 mRNA levels were predicted to increase as Atf3 protein is lost ., We combined the two systems for ITF and miRNA inhibition of Atf3 to investigate whether an ITF may influence the profiles in the presence of miRNA ( s ) ( Figure 6A ) ., Profiles for total Atf3 mRNA ( Figure 6B ) , Atf3 protein ( Figure 6C ) and Egr1 mRNA ( Figure 6D ) were similar to those for model extension 4 ( Figure 5 , C–E ) indicating that miRNAs alone could suffice to switch off Atf3 mRNA and protein expression ., In considering mechanisms of regulation of IEG expression , the emphasis is usually on the signals and transcriptional regulators required for upregulation ., Here , we explored possible mechanisms associated with downregulation of IEG expression and termination of the response ., We focused on Atf3 as an IEG in which there is great interest because of its association with inflammation , cancer and cardiac dysfunction 11–14 ., The report that Atf3 can bind to its own promoter to inhibit transcription 22 is , at first sight , attractive ., The experimental approaches used remain the most appropriate , but all such studies are potentially subject to artefacts resulting from overexpression or elimination of a single transcription factor in isolation ., Developing a mathematical model provides an alternative means of testing the hypothesis ., In this case , the model demonstrated that such a negative feedback loop is implausible in the context of our cardiomyocyte system ( Figure 2 ) ., This raises the question of whether any such self-regulating transcriptional system can operate ., For an inhibitory TF , the model prediction is that a redundant feedback loop is most likely to be generated in which the TF serves to regulate only its own expression , restricting the possibility that it accumulates to a sufficient degree to operate in an efficient inhibitory loop with another gene ., The scenario would differ in the case of a positive TF driving expression of second phase genes ., Here , self-regulation can provide a means of moderating downstream gene expression ., Our experimental data suggest that , whereas ERK1/2 themselves promote upregulation of Atf3 , downstream activation of RSKs ( or other BI-D1870-sensitive kinases ) is potentially important in terminating Atf3 mRNA and , presumably , protein expression ( Figure 3 , A and B ) ., This resulted in the development of three further extensions to the model ., Whilst we have included biologically appropriate mechanisms at each model revision one outcome of this has been an increase in the signalling cascade complexity ., This in itself has led to a more robust signalling response ., The initial concept , that delayed activation of RSKs may result in delayed activation of a pre-existing inhibitory TF for signal termination , appeared attractive ., However , model extension 2 demonstrated that direct activation of the putative RSK-responsive RTF was almost simultaneous with activation of the ERK1/2-responsive TF and simply titrated down the levels of Atf3 mRNA rather than altering the profile ., Thus , the delay between activation of ERK1/2 and activation of RSKs is insufficient for this scenario to be viable and a greater delay is required before initiation of the negative signal ., Interestingly , since a delay is necessary , this also negates the possibility of other acute events ( e . g . histone deacetylation ) being responsible for terminating Atf3 mRNA expression ., The timing strongly suggests that a second transcriptional event is required , either to produce an alternative inhibitory TF ( termed ITF , model extension 3 ) and/or one or more miRNAs ( model extension 4 ) ., Here , developing first model extensions 3 and 4 independently and then combining the two strongly suggested that an ITF would not affect Atf3 mRNA and protein expression whereas miRNA ( s ) most probably would suffice to generate the profiles seen experimentally ., Interestingly , the miRNA effect was particularly dominant on the Atf3 protein profile ., The model also demonstrated that a modest increase in the rate of degradation of Atf3 mRNA by the miRNA within the range predicted by other computational models 35 , 36 could suffice for the Atf3 mRNA profiles to approximate to the experimental data ., Further studies of cardiomyocyte miRNAs are clearly required ., Neonatal rat cardiomyocytes were prepared as previously described 37 and exposed to 2 µM PD184352 ( to inhibit the ERK1/2 cascade 38 ) or 10 µM BI-D1870 ( to inhibit RSKs 31 ) for 70 min , to 100 nM ET-1 for 1 h or to inhibitor for 10 min prior to addition of ET-1 ( 1 h ) ., PD184352 and BI-D1870 were from Enzo Life Sciences and were dissolved in DMSO ., ET-1 was from Bachem UK ., RNA was extracted and quantitative PCR performed as described in 10 ., Affymetrix microarray data for cardiomyocytes exposed to ET-1 in the absence or presence of PD184352 or BI-D1870 were published in 9 ., The data are available from ArrayExpress ( accession nos . E-MIMR-3 , E-MIMR-37 , E-MEXP-3393 , E-MEXP-3394 , E-MEXP-3678 , E-MEXP-3679 ) ., Probeset sequences that were significantly upregulated and not assigned to a protein-coding gene were analysed by BLAST search against the rat and mouse genomes ( www . ncbi . nlm . nih . gov/genome/seq/BlastGen/BlastGen . cgi ? taxid=10116; www . ncbi . nlm . nih . gov/genome/seq/BlastGen/BlastGen . cgi ? taxid=10090; cross-species megaBLAST ) to identify those which recognise miRNAs ., Data analysis and curve-fitting to experimental data used GraphPad Prism ., Full details of the mathematical modelling are provided in Text S1 . | Introduction, Results, Discussion, Materials and Methods | Activating transcription factor 3 ( Atf3 ) is rapidly and transiently upregulated in numerous systems , and is associated with various disease states ., Atf3 is required for negative feedback regulation of other genes , but is itself subject to negative feedback regulation possibly by autorepression ., In cardiomyocytes , Atf3 and Egr1 mRNAs are upregulated via ERK1/2 signalling and Atf3 suppresses Egr1 expression ., We previously developed a mathematical model for the Atf3-Egr1 system ., Here , we adjusted and extended the model to explore mechanisms of Atf3 feedback regulation ., Introduction of an autorepressive loop for Atf3 tuned down its expression and inhibition of Egr1 was lost , demonstrating that negative feedback regulation of Atf3 by Atf3 itself is implausible in this context ., Experimentally , signals downstream from ERK1/2 suppress Atf3 expression ., Mathematical modelling indicated that this cannot occur by phosphorylation of pre-existing inhibitory transcriptional regulators because the time delay is too short ., De novo synthesis of an inhibitory transcription factor ( ITF ) with a high affinity for the Atf3 promoter could suppress Atf3 expression , but ( as with the Atf3 autorepression loop ) inhibition of Egr1 was lost ., Developing the model to include newly-synthesised miRNAs very efficiently terminated Atf3 protein expression and , with a 4-fold increase in the rate of degradation of mRNA from the mRNA/miRNA complex , profiles for Atf3 mRNA , Atf3 protein and Egr1 mRNA approximated to the experimental data ., Combining the ITF model with that of the miRNA did not improve the profiles suggesting that miRNAs are likely to play a dominant role in switching off Atf3 expression post-induction . | Activating transcription factor 3 ( Atf3 ) is an important regulatory transcription factor which is associated with inflammation , restraint of the immune response and cancer ., In this work , we develop a series of mathematical models to understand how Atf3 may be regulated ., Informed with data from the literature and our own experiments , we show that self-regulation of Atf3 does not allow for variation between experimentally observed Atf3 mRNA and Atf3 protein expression profiles ., A fast-acting signal via phosphorylated RSK is also shown to be implausible for similar reasons ., Extending our mathematical model further , we postulate for the first time , that the observed dynamical variation in Atf3 mRNA and protein can be described by microRNAs downstream of RSKs ., The further inclusion of an inhibitory transcription factor for Atf3 expression has little effect on these findings . | systems biology, mathematics, cell biology, calculus, applied mathematics, biology and life sciences, physical sciences, molecular cell biology, differential equations | null |
journal.pcbi.1002501 | 2,012 | Are Long-Range Structural Correlations Behind the Aggregration Phenomena of Polyglutamine Diseases? | Polyglutamine ( polyQ ) diseases involve a set of nine late-onset progressive neurodegenerative diseases caused by the expansion of CAG triplet sequence repeats 1 ., These repeats result in the transcription of proteins with abnormally long polyQ inserts ., When these inserts expand beyond a normal repeat length , the affected proteins form toxic aggregates 2 leading to neuronal death ., PolyQ aggregation takes place through a complex multistage process involving transient and metastable structures that occur before , or simultaneously , with fibril formation 3–9 ., Experimental findings suggest that the therapeutic target for polyQ diseases should be the soluble oligomeric intermediates , or the conformational transitions that lead to them 9 , 10 , and not the insoluble ordered fibrils ., These findings , common to all amyloid diseases 11 , have spurred efforts to understand the structural attributes of soluble oligomers and amyloidogenic precursors ., The free energy landscapes of polyQ aggregates display countless minima of similar depth that correspond to a great variety of metastable and/or glassy states ., The aggregation kinetics of pure polyQ have been described as a nucleation-growth polymerization process 4–6 , 12 , where soluble expanded glutamine requires a considerable time lag for the creation of a critical nucleus , which then readily converts into a sheet in the presence of a template 13 ., However , the “time lag” seems to properly be associated with the formation of the fully aggregated precipitates , since soluble aggregates – sometimes called “protofibrils” – that form during the putative lag phase have been reported 14 , 15 ., The variety of polyQ soluble and insoluble aggregates might correlate with the conformational flexibility of monomeric ( non-aggregate single-chain ) polyQ regions , which are influenced by the conformations of neighboring protein regions 4 , 16–18 ., One striking example of this conformational wealth – and still a source of controversy– is given by the polyQ expansion in the N-terminal of the huntingtin protein that is encoded in the exon 1 ( EX1 ) of the gene ., The N-terminal amino acid sequence consists of a seventeen , mixed residue sequence , the polyQ region of variable length , two polyproline regions of 11 and 10 residues separated by a region of mixed residues , and a C-terminal sequence ., Toxicity develops after the polyQ expansion exceeds a threshold of approximately 36 repeats , leading to Huntingtons disease ., The flanking sequences have been shown to play a structural role in polyQ sequences , both in synthetic and natural peptides , and both in monomeric or aggregate form 4 , 16 , 17 , 19 ., In particular , a polyproline ( polyP ) region immediately adjacent to the C-terminal of a polyQ region has been shown to affect the conformation of the polyQ region; the resulting conformations depend on the lengths of both the polyQ and polyP sequences 16 , 17 , 20 , 21 ., In this work , we set out to obtain a conceptual and quantitative understanding of the role played by a polyP sequence that is placed at the C-terminal of a polyQ peptide , which is relevant for the understanding of the behavior of the EX1 segment in the huntingtin protein ., Sedimentation aggregation kinetics experiments 17 show that the introduction of a sequence C-terminal to polyQ in synthetic peptides decreases both the rate of formation and the apparent stability of the associated aggregates ., The polyP sequence can be trimmed to without altering the suppression effect , but a sequence is ineffective ., There are no effects when the polyP sequences are attached to the N-terminal or via a side-chain tether 17 ., These experiments were complemented with CD spectra for monomeric peptides , where the presence of polyP at the C-terminal of showed remarkable changes in the spectra ., Analysis of their data led the authors to propose that addition of the C-terminal sequence does not alter the aggregation mechanism , which is nuclefated growth by monomer addition with a critical nucleus of 1 monomer ( for ) , but destabilizes both the -helical and the ( still unknown ) aggregation-competent conformations of the monomer ., These experimental results were unexpected: although a single proline residue interrupting an amyloidogenic sequence can decrease the propensity of that sequence to aggregate 22 , 23 , Pro replacements in amyloidogenic sequences placed in turns or disordered regions do not alter the aggregate core 23 ., Here , we consider monomeric polyQ and polyQ-polyP chains , and quantify changes brought about in the conformations of the polyQ sequences by the addition of the polyP sequences at their C-terminal ., In order to assess these changes , one must first characterize the conformation of pure monomeric polyQ in water ., Wildly diverse conformations have been postulated experimentally for monomeric polyQ , including a totally random coil , -sheet , -helix , and PPII structures ., At present there is growing experimental evidence that single polyQ chains are mainly disordered 6 , 13–15 ., The solvated polyQ disorder , however , is different from a total random coil or a protein denatured state ., In particular , atomic X-ray experiments 18 show that single chains of polyQ ( in the presence of flanking sequences ) present isolated elements of -helix , random coil and extended loop ., Single-molecule force-clamp techniques were used to probe the mechanical behavior of polyQ chains of varying lengths spanning normal and diseased polyQ expansions 24 ., Under the application of force , no extension was observed for any of the polyQ constructs ., Further analysis led the authors to propose that polyQ chains collapse to form a heterogeneous ensemble of globular conformations that are mechanically stable ., Simulations results for the monomer conformation have also been contradictory 25–31 ., It is interesting that in the search for soluble prefibrillar intermediates , an -sheet was proposed to play a role in polyQ toxicity 32 , 33 ., In these molecular dynamics simulations , polyQ monomers of various lengths were found to display transient -strands of four residues or less ., The authors proposed that fibril formation in polyQ may proceed through strands intermediates 33 ., More recently , a molecular dynamics study of hexamers of in explicit water showed that -sheet aggregates are very stable ( more stable than -sheets ) 34 ., These results strongly support the idea that -sheet may either be a stable , a metastable , or at least a long-lived transient , secondary structure of polyQ aggregates ., Coming back to the monomeric polyQ conformation , further simulation evidence 35–38 supports the experimental findings that monomeric polyglutamine of various lengths is a disordered statistical coil in solution ., The disorder is inherently different from that of denatured proteins and the average compactness and magnitude of conformational fluctuations increase with chain length 35 ., In addition , the coils may present considerable -helical content 38 , but there are acute entropic bottlenecks for the formation of -sheets ., The molecular dynamics results presented here for single polyQ and polyQ-PolyP chains consisting of , , , , , and glutamine residues are in qualitative agreement with the experimental and simulation results mentioned above: polyQ is primarily disordered , with non-negligible -helical content and a small population of other secondary structures including both and strands ., The addition of polyP reduces the population of the region of Ramachandran plot 39 , and increases the population of and PPII Ramachandran regions for all PolyQ lengths ., If one considers secondary structure motifs ( i . e . , hydrogen-bonds patterns in addition to dihedral angles ) , the addition of the polyP segment increases the populations of the PPII helices and turns , and decreases the -helical content of all peptides but ( which may have a protective effect against aggregation , as discussed later ) ., The addition of polyP does not change the average radius of gyration of polyQ , but changes the radius of gyration distribution function for , that becomes dependent on the prolyl bond isomerization state ., Most importantly , the addition of polyP decreases the population of small and strands , and and hairpins ., Since the extended strands and hairpins in both and forms are found only in a small fraction of the structures , we used a novel statistical measure based on the odds ratio construction 40 to quantify to study the secondary structural propensities 41 , 42 , thereby learning about the possibility of the growth of such secondary structures under nucleation conditions ., This study , also supported by more conventional linear correlation analysis , provides evidence that among all the peptides studied here , only exhibits a long-range correlation between all glutamine residue pairs that favors formation of both and -strands ., This correlation is suppressed by the addition of only six proline residues to the C-terminal of the peptide , which suggests a mechanism in which nucleation starts at these scarcely populated secondary structures ( mainly , , and strands , as well as -hairpins and -hairpins ) and can only spread through positive correlations in polyQ peptides of approximately 40 residues or longer ., This paper is organized as follows ., The Methods section details our simulation methodology and analysis ., Specifically , we discuss the generalized Replica Exchange scheme used here for enhanced sampling , the simulation details , our clustering techniques to identify the Ramachandran regions and the secondary structural motifs , and the odds ratio construction , used here to study the correlations between residues ., In the Results section , we present our results with a focus on a statistical analysis of the equilibrium conformations based on, ( i ) Ramachandran regions, ( ii ) secondary structure, ( iii ) correlation analysis and, ( iv ) radius of gyration ., A discussion of our results and a short summary of this work is given in the last section ., Room temperature , regular molecular dynamics ( MD ) simulations are often too computationally limited to carry out a full sampling of the conformational space of a biomolecular system and generate a reliable statistical ensemble ., Thus , in order to deal with the sampling issue , we make use of a replica exchange scheme 43 , 45 ., In the replica exhange molecular dynamics ( REMD ) 43 , 46 method , one considers several replicas of a system subject to some sort of ergodic dynamics based on different Hamiltonians , and attempts to exchange the trajectories of these replicas at a predetermined rate to increase the barrier crossing rates ( i . e . , decrease the ergodic time scale ) ., One possibility is to successively increase the temperatures of the replicas 46 ., This method , known as parallel tempering , is here referred to as Temperature REMD ( T-REMD ) ., Another possibility 43 is to construct the replicas by adding a biasing potential to the original Hamiltonian that acts on some collective variable that describes the slow modes of the system that need “acceleration” ., This method can be referred to as Hamiltonian REMD ( H-REMD ) ., In practice , T-REMD is used to promote the barrier crossing events in a generic way but the use of H-REMD allows one to directly focus on specific slow modes of the system , such as the cis-trans isomerization of proline amino acids which involves a barrier of 10 to 20 Kcal/mol 47 ., A combination of the two methods , known as Hamiltonian-Temperature REMD ( HT-REMD ) 41–44 provides for a practical way to reduce the computational costs associated with REMD sampling , since it facilitates the sampling by both means ., In this work , we used the T-REMD and HT-REMD methods for polyQ and polyQ-polyP peptides , respectively ., In the T-REMD method , one replica runs at room temperature and the rest of the replicas run at higher temperatures ., Care must be taken with respect to the choice of the number of replicas and their temperatures ., The performance of the setting can be checked by monitoring the exchange rate between the neighboring replicas ( i . e . , with closest temperatures ) as well as the ergodic time scale of the “hottest” replica ., The equilibrium conformational ensemble is then generated by taking the structures at a predetermined rate from the trajectory of the replica at the lowest ( room ) temperature ., In the HT-REMD method , the replicas have different biasing potentials ., The biasing potential is usually described in terms of a collective variable , defined as a smooth function of the atomic positions ., The corresponding free energy or potential of mean force ( PMF ) 48 , ( where the angular brackets denote the equilibrium ensemble average ) , provides for an ideal biasing potential ., Indeed , if the biasing potential is exactly , then the probabilities of different values of the collective variable would all be equal , since there are no barriers present ., Although the true free energy is typically unknown in advance , a roughly approximate is often sufficient to improve the sampling considerably in an H-REMD or HT-REMD setting ., Such free energies can be computed in a variety of ways 48 ., For the polyQ-polyP systems , some of the slow modes originate in the cis-trans isomerization of the prolyl bonds , that occur when polyproline is in solution ., We have recently carried out extensive work on proline-rich systems 41 , 42 , 44 , 47 , 49 and can take advantage of the free energy profiles previously obtained for polyproline of various lengths 44 , calculated using the Adaptively Biased Molecular Dynamics ( ABMD ) 50 , 51 method ., The ABMD method is an umbrella sampling method with a time-dependent biasing potential , which can be used in conjunction with the REMD protocol , by combining different collective variables and/or temperatures on a per-replica basis 43 , 50 ., Currently , the ABMD method has been implemented into the AMBER v . 10 , 11 simulation package 52 ., Details of the calculation of the polyproline potentials are given elsewhere 41 , 42 , 44 , 47 ., The HT-REMD simulations proceeded in several stages ., We recycled the previously computed free energies associated with a collective variable that “captures” the cis-trans transitions of the prolyl bonds of polyproline peptides of different lengths in implicit water at different temperatures ., The collective variable used for these calculations is defined based on the backbone dihedral angle of prolyl bonds , ( here sum runs over all the prolyl bonds ) ., The dihedral angle takes the values around and for cis and trans conformations , therefore can “capture” different patterns of the cis/trans conformations in any proline-containg peptide ., The biasing potentials , transfered from our previous calculations were then refined for the polyQ-polyP peptides using similar simulation settings ., Next , several additional replicas running at the lowest temperature were introduced into the setup ., One of these replicas is completely unbiased , and therefore samples the Boltzmann distribution at ., The other replicas , also at , are subject to a reduced bias ( i . e . , these biasing potentials are scaled down by a constant factor ) ., The purpose of these “proxy” replicas is to ensure adequate exchange rates between the conformations , and thereby enhance the mixing 43 ., Data was then taken from the unbiased replica at a suitable , predetermined rate ., Simulations were carried out for the peptides with sequence ( denoted as ) and ( denoted as ) ., These peptides include , , , , , , , , , , and ., In each case , we refer to the glutamine and proline residues as and , respectively ., The simulations were carried out using the AMBER 52 simulation package with the ff99SB version of the Cornell et al force field 53 with an implicit water model based on the Generalized Born approximation ( GB ) 54 , 55 including the surface area contributions computed using the LCPO model 56 ( GB/SA ) ., For more simulation details , our implementation of the REMD scheme and a discussion of convergence issues , please see the Supporting Information ( Text S1 ) ., We used the ( , ) dihedral angles ( see Fig . 1 for their definition ) to identify different regions 57 of the Ramachandran map 39 ., Table 1 provides the corresponding definition for these regions ., Although this delineates clear regions for the dihedrals of most residues , it turns out that the populations may overlap around the borders ., In order to handle this situation , we used a clustering technique as explained in the Supporting Information ( Text S1 ) to classify the conformations , rather than strictly enforcing the sharp boundaries between the defined regions ., Although the backbone dihedral angles of all the residues forming a right-handed -helix fall into the region of Ramachandran map , many of the residues in this region do not actually form -helices ., As a matter of fact , several other secondary structural motifs , such as and helices as well as random coil and turn are characterized by or may involve backbone dihedral angles falling in the same region ., An interesting example is provided by polyglutamine itself ., It has been suggested recently 32–34 that an -sheet , whose backbone dihedral angles alternate between the and helical regions , can be a stable , metastable , or at least a long-lived transient secondary structure in oligomers ., In general , for a residue to be considered to belong to a given secondary structure , it is not enough to identify the Ramachandran region of its dihedral angles ., Thus , we used the secondary structure prediction program DSSP 58 , 59 that uses not only the backbone diheral angles , but also the inter-residual hydrogen bonding as well as the relative position of the C atoms to identify secondary structural motifs ., For our peptides , the DSSP secondary structures with highest probabilities were:, ( i ) helices , including and types ,, ( ii ) turns , including H-bonded turns and bends ,, ( iii ) coils ., There are also isolated residues involved in bridges and extended strands , participating in the ladders with small probabilities ., Since DSSP does not specifically identify isolated or strands ( i . e . , strands not H-bonded to another strand of their type ) or hairpins , we used a combination of H-bonding results from DSSP analysis and the Ramachandran regions from the clustering analysis to define and strands and hairpins ., A strand is defined here as at least adjacent residues all falling into the region of Ramachandran plot ., A strand is referred to as isolated if none of its residues is H-bonded ., A hairpin is defined as two adjacent strands with a turn in between and at least one H-bond between the two strands ., The turn between the two strands of a hairpin could be H-bonded or not and is of any length but it has to have the geometrical form of a turn , ( i . e . , identified as bend by DSSP ) ., Each of the two strands has at least three adjacent residues in region to ensure the structure is relatively extended ., At least one of these three residues are H-bonded to another residue in the other strand ., We define an repeat as two adjacent residues , whose backbone dihedral angles alternate between and regardless of the order ( i . e . , this includes both and ) ., An strand is formed from adjacent residues , involving alternating and repeats ., In this definition , an strand is either or and an strand is either or but not ., An isolated strand is defined as an strand not H-bonded to another strand , and the hairpin is defined as two adjacent strands with a turn in between and at least one H-bond between the two strands , similar to the hairpin ., Another relatively extended secondary structure is PPII that is defined here as adjacent residues whose dihedral angles fall into the PPII region of Ramachandran plot ., A PPII structure , is defined as a structure having adjacent PPII residues ., A summary of these secondary structures is given in Table 1 ., Finally , we determined the type of turn from both the DSSP analysis and our Ramachandran region clustering analysis ., DSSP distinguishes between H-bonded turns and geometrical bends that do not involve any H-bonding ., The DSSP analysis can be also used to identify and types based on the number of residues involved , which is 4 and 3 respectively ., The dihedral angles of the two middle residues of turns ( i . e . the second and the third residues ) can be used to partition turns into more types such as I , I′ , II , II′ , etc . but we will only consider type I- that involves an sequence and the “other” type turns that involve other combinations of dihedral angles ., Since the population of “other” combinations is relative small , we group these all together ., To quantify how the secondary structures of Gln residues influence each other we made use of the odds ratio ( OR ) construction 40–42 ., The OR is a descriptive statistic that measures the strength of association , or non-independence , between two binary values ., The OR is defined for two binary random variables ( denoted as and ) as: ( 1 ) where is the joint probability of the event ( with and taking on binary values of 0 and 1 ) ., For the purposes of this study , we can think of and as being some characteristic properties describing the conformations of different residues ., For example , the variables could be assigned values of 1 or 0 depending on whether the backbone dihedral angles of corresponding residue falls into the region of Ramachandran plot or not ., We denote this definition of OR as OR ., Similarly one can define based on the involvement of residues in repeats ., In this case , to define the of two given residues and , the probabilities are defined such that the variables and take the values 1 or 0 depending on whether or not the corresponding residue is involved in an repeat as defined in the last subsection ., For instance , if and only if residue either is in the region and is neighboring a residue in the region , or it is in the region and is neighboring a residue in the region ., Note that in general , to calculate the of two residues , dihedral angles of not only the two residues but also their neighbors are needed , i . e . , up to 6 residues could be involved ., The usefulness of the OR in quantifying the influence of one binary random variable upon another can be readily seen ., If the two variables are statistically independent , then so that ., In the opposite extreme case of ( complete dependence ) both and are zero , and the OR is infinite ., Similarly , for rendering ., To summarize , an OR of unity indicates that the values of are equally likely for both values of ( i . e . , , and are therefore independent ) ; an OR greater than unity indicates that is more likely when ( and are positively correlated ) , while an OR less than unity indicates that is more likely when ( and are negatively correlated ) ., It is convenient to recast the log of the OR in terms of free energy language ., If one expresses the probability of the events in terms of a free energy : ( 2 ) then the ratio of probabilities translates into a free energy difference: ( 3 ) Clearly , the logarithm of the OR then maps onto the difference of those differences , i . e . , ( 4 ) For the case of statistically independent properties , ; otherwise , this quantity takes on either positive or negative values , whose magnitude depends on the mutual dependence between the two variables ., The standard error in its asymptotic approximation is: ( 5 ) in which is the total number of independent events sampled ., While this development may be perceived as purely formal , the use of an OR analysis couched in terms of free energy language provides for a useful and intuitive measure of the inter-residual correlations , as has been illustrated before 41 , 42 ., In this work , our OR-based correlation analysis is supported by the conventional linear correlation analysis ., We have used the correlation coefficient ( also know as cross-correlation or Pearson correlation ) of dihedral angles of glutamine residues to measure the correlation of glutamine residues in different situations ., We emphasize that in the context of secondary structural propensities , the odds ratio analysis is more powerful than the correlation coefficient since it eliminates the noise associated with the dihedral angles ., This noise may dominate the linear correlation results such that even substantial correlations may be completely ignored ., The OR-based correlation analysis , combined with the clustering technique explained here takes into account both nonlinearity and multivariate components of amino acid correlations in a peptide chain , although in some particular cases a conventional univariate linear correlation may reveal a correlation as we will report in the results ., In the context of this paper , the multivariate component is particularly evident when the correlation of repeats is considered , since this may involve and angles of up to six residues for each single odds ratio calculation ., Figure 1b shows the Ramachandran plot of a typical glutamine residue , for which the clusters in the different regions are computed according to the protocol described in the Methods section ., Four clusters can be identified in these plots including PPII ( blue ) , ( yellow ) , ( gray ) , and ( pink ) ., Figures S2 and S3 show the Ramachandran plots of all 40 glutamine residues of both and ., Considering these , as well as similar plots for other peptides ( not shown here ) , we observe the following trends:, ( i ) The dominant region of most residues is the cluster that is present in all residues , except for the glutamines immediately followed by a proline , for which this region is precluded;, ( ii ) PPII and clusters are present in almost all residues;, ( iii ) The cluster is present in more than half of the residues but its population is often very small;, ( iv ) Compared to , displays regions with higher non- intensities , particularly for the cluster ( see , , , and ) ., Figure 2 plots the percent population of the , PPII , and regions of glutamine residues ( top , middle and bottom rows , respectively ) in terms of the residue number ., The left column shows results for red and blue and the right column for red and blue ., Table 2 presents the population of the different Ramachandran regions ( averaged over all glutamine residues ) and the repeats , the secondary structure motifs , and the “extended structures” including hairpins ., The residue populations in the Ramachandran plot show that , on average , 67–87 of the residues are in the region of the Ramachandran plot , 5–13 of the residues are in the PPII region and 5–17 of the residues are in the region ., The PPII and regions are almost always equally probable , as can be seen in Figs ., 2 , S2 , S3 ., The lowest population belongs to the region , comprising only 3–6 although in certain residues it could be as high as 38% as , for instance , in in where the content of correlates with the presence of turns ., The addition of P decreases the population of the Ramachandran region and increases that of the and PPII regions , while leaving the small population of approximately invariant ., In peptides , proline residues are excluded from the statistical analysis so that only Q residue propensities are compared ( for instance , when we state that the average helical content of is 43% , it means that 43% of all Q residues are in a helix – the P residues are not counted in the statistic ) ., Figure 2 shows that the populations of the PPII and regions are always higher at the two ends of the polyQ peptides , particularly at the C-terminal ., When a short proline segment is added at the C-terminal of polyQ , the population of these regions in the neighboring glutamines increases even more ., For peptides shorter than ( not shown here ) , the population of the PPII- region decreases in the middle of the peptide , but for ( red line ) we see a small peak in the middle of the peptide for both PPII and regions ., In , we have two small peaks ( rather than a single peak ) centered around residues 13 and 25 for both the and PPII regions ., The presence of the prolines at the C-terminal of a polyglutamine can drastically alter the population distribution ., Fig . 2 shows that the few relatively wide peaks of the -PPII regions in both and are replaced by several narrow peaks of larger heights ., Regarding the residues involved in repeats , one can see from Fig . 2e , f that the distribution of these repeats throughout these peptides depends both on the position of glutamine residues and the presence or absence of the C-terminal prolines although , as seen in Table 2 , the average content is similar ( 6–7% ) in all four peptides: , , , and ., We note that the distribution of content in the peptide is mostly determined by the content as the content is abundant in these peptides and most residues are involved in an repeat ., One can compare Fig . 2e , f with Figs ., S2 , S3 and observe similar behaviour , i . e . , the residues with high content ( Fig . 2e , f ) have more intense clusters ( pink clusters in Figs . S2 , S3 ) ., When one considers not only the backbone dihedral angles i . e . , the ( , ) regions occupied by individual glutamine residues , but also the inter-residual hydrogen bonding and the relative positions of the atoms , one can identify different secondary structures , particularly -helical segments in many of the sampled conformations ., Short helices are also possible but the majority of the residues are either in a turn or a coil conformations according to both DSSP 58 , 59 and STRIDE 60 analysis ., Figure 3 plots the helical , turn , and coil content of the individual glutamine residues against their residue numbers for , , , and ., Figure 4 shows plots of select conformations of and peptides , as generated by VMD 61 using STRIDE 60 for the secondary structure assignment ., Table 2 lists the population of helix , turn , and “other” secondary structures as obtained from DSSP , averaged over all residues ., The “other” secondary structure category includes mainly what DSSP identifies as “loop or irregular” – sometimes called “coil” in other programs – but which may also include a very small population of other secondary structures such as extended strand and “isolated -bridge” ., We use the protocols explained in Methods section to further identify these , as well as other extended structures ( Tables 2 and 3 ) ., When the population of residues in the region is compared to the actual helical content , one realizes that the majority of the residues in the region do not form or any other type of helices ., Many of these residues in the region are followed and/or preceded by a residue in a different Ramachandran region , such as , as discussed in the previous subsection , forming an repeat ., Similarly an repeat does not necessarily form an strand ., Table 2 gives the population of the structures ( or conformations ) having at least one segment in one of the extended conformation forms , as defined in Methods section , including and strands either in the isolated form of length 3 ( or length 4 in parenthesis ) or in the hairpin form as well as PPII structures of length 3 ( or length 4 in parenthesis ) ., Note that unlike the other populations in part, ( a ) and, ( b ) in Table 2 , the population of extended secondary structures in part, ( c ) is not averaged over the residues ., Instead , we counted all the conformations having at least one such secondary structures in the polyQ portion of the molecule and divided this number by the total number of sampled conformations ., These structures are less common than helices or turns , but they are possible and form a small subpopulation of the secondary structures . | Introduction, Methods, Results, Discussion | We have characterized the conformational ensembles of polyglutamine peptides of various lengths ( ranging from to ) , both with and without the presence of a C-terminal polyproline hexapeptide ., For this , we used state-of-the-art molecular dynamics simulations combined with a novel statistical analysis to characterize the various properties of the backbone dihedral angles and secondary structural motifs of the glutamine residues ., For ( i . e . , just above the pathological length for Huntingtons disease ) , the equilibrium conformations of the monomer consist primarily of disordered , compact structures with non-negligible -helical and turn content ., We also observed a relatively small population of extended structures suitable for forming aggregates including - and -strands , and - and -hairpins ., Most importantly , for we find that there exists a long-range correlation ( ranging for at least residues ) among the backbone dihedral angles of the Q residues ., For polyglutamine peptides below the pathological length , the population of the extended strands and hairpins is considerably smaller , and the correlations are short-range ( at most residues apart ) ., Adding a C-terminal hexaproline to suppresses both the population of these rare motifs and the long-range correlation of the dihedral angles ., We argue that the long-range correlation of the polyglutamine homopeptide , along with the presence of these rare motifs , could be responsible for its aggregation phenomena . | Nine neurodegenerative diseases are caused by polyglutamine ( polyQ ) expansions greater than a given threshold in proteins with little or no homology except for the polyQ regions ., The diseases all share a common feature: the formation of polyQ aggregates and eventual neuronal death ., Using molecular dynamics simulations , we have explored the conformations of polyQ peptides ., Results indicate that for peptides ( i . e . , just above the pathological length for Hungtingtons disease ) , the equilibrium conformations were found to consist primarily of disordered , compact structures with a non-negligible -helical and turn content ., We also observed a small population of extended structures suitable for forming aggregates ., For peptides below the pathological length , the population of these structures was found to be considerably lower ., For longer peptides , we found evidence for long-range correlations among the dihedral angles ., This correlation turns out to be short-range for the smaller polyQ peptides , and is suppressed ( along with the extended structural motifs ) when a C-terminal polyproline tail is added to the peptides ., We believe that the existence of these long-range correlations in above-threshold polyQ peptides , along with the presence of rare motifs , could be responsible for the experimentally observed aggregation phenomena associated with polyQ diseases . | physics, chemistry, chemical biology, chemical physics, biophysics | null |
journal.ppat.1001169 | 2,010 | Strain-Specific Differences in the Genetic Control of Two Closely Related Mycobacteria | The primary cause of tuberculosis is the human pathogenic bacterium Mycobacterium tuberculosis ., The host cells of M . tuberculosis are macrophages and the bacilli have developed numerous adaptations to survive within these powerful immune effector cells ., For example , human pathogenic strains of M . tuberculosis inactivate microbicidal superoxide via katalase 1 , avoid the detrimental effects of iNOS products 2 , skew the anti-mycobacterial response in macrophages towards production of anti-inflammatory molecules 3 , 4 , and favour necrosis over apoptosis 5 , 6 , 7 ., Interestingly , circulating strains of M . tuberculosis may differ in their pathogenic potential 8 , 9 ., Since humans and M . tuberculosis have co-evolved over millennia , a question remains if and to what extent M . tuberculosis has adapted to genetically distinct hosts ., Indeed , two studies conducted in ethnically mixed samples detected a non-random association of M . tuberculosis strains with distinct ethnic populations 10 , 11 ., These observations are supported by the results of several genetic association studies that detected preferential associations between a Toll-like receptor 2 ( TLR2 ) polymorphism and tuberculosis meningitis caused by Beijing strains 12 , as well as between variants of 5′-lipoxygenase ( ALOX5 ) and pulmonary tuberculosis caused by M . africanum , but not M . tuberculosis 13 ., In addition , variants of the immunity-related GTPase M ( IRGM ) were associated with protection from pulmonary tuberculosis due to Euro-American strains of M . tuberculosis 14 ., Due to the complex interactions of M . tuberculosis and humans in exposed populations , it is possible that those results may have been confounded by unrecognized factors ., In the absence of independent replication studies , the question of strain specific genetic effects as a consequence of M . tuberculosis human co-evolution still awaits testing under carefully controlled conditions ., M . bovis Bacille Calmette-Guerin ( BCG ) strains are phylogenetic descendants of an ancestral BCG stock originally derived from virulent M . bovis through in vitro propagation 15 , 16 , 17 ., Attenuation of the original BCG stock occurred as a result of deletions in the M . bovis genome , specifically the region of difference 1 ( RD1 ) 18 , 19 ., Loss of RD1 is common across all BCG strains , although additional genetic alterations have been identified for each strain ., BCG Russia and BCG Pasteur are among the most phylogenetically distant BCG strains 15 ., Genetic events identified in BCG Russia include the deletion of RD Russia ( Rv3698 ) 20 , an insertion mutation in the recA gene ( recA_D140* ) 21 , and the presence of an IS6110 element in the promoter region of the phoP gene 15 , 22 ., BCG Pasteur is characterized by the loss of RD2 , nRD18 , and RD14 23 , 24 , 25 as well as a number of single point mutations and duplication events 22 , 23 , 26 , 27 ., Phenotypic differences between BCG Pasteur and BCG Russia can therefore be tentatively linked to these known changes in gene content and an unknown number of point mutations ., A number of unresolved questions surround the BCG host interplay which is characterized by highly variable host responsiveness ., For example , the immunogenicity of the same strain of BCG given to vaccinees of different genetic background can vary tremendously 28 , 29 while host responses triggered by different strains of BCG are equally divergent 30 ., On a population scale , BCG strains differ in the adverse reactions they trigger 31 and there is evidence that the protective effect of BCG vaccination against tuberculosis meningitis varies among ethnically divergent population groups 32 ., Taken together , these data suggest that , similar to tuberculosis susceptibility , host responsiveness may reflect specific host-BCG strain interactions ., To test this possibility , we compared the genetic control of closely related strains of BCG in a mouse model of infection ., Recombinant congenic ( RC ) strains are a set of genetically related inbred strains ., In RC strains , discrete chromosomal segments of donor genome ( 12 . 5% ) are transferred onto a recipient genetic background ( 87 . 5% ) through a double backcross and corresponding strains are derived by subsequent inbreeding 33 ., The AcB/BcA panel used in the present study was derived from a reciprocal double backcross between C57BL/6J and A/J 34 , two mouse strains known to differ in their susceptibility to M . bovis BCG strain Montreal 35 ., Each RC strain is genetically distinct with its own unique genome ., The genomes of all RC strains have been mapped extensively and represent frozen replicas of recombinant progenitor genomes with known genomic boundaries of chromosomal segments derived from the two progenitor strains ., A major advantage of RC strains over conventional crosses is that any phenotype can be measured repeatedly in genetically identical mice of a RC strain , greatly improving the accuracy of the phenotypic estimates ., In the present study , 35 distinct AcB/BcA strains were infected with a low dose of either BCG Pasteur or BCG Russia ., A genetic analysis of the bacillary counts in the spleen and lungs of these strains identified general , as well as tissue- and BCG strain-specific susceptibility loci for BCG infection ., These results demonstrated that the host response to mycobacteria reflects a genetically controlled , joint effect of both host and pathogen ., Our findings established strain specific effects of the host-mycobacteria interplay in the absence of selective pressure and , therefore , argue in favour of additional host-mycobacterial adaptation during the co-evolution of humans and mycobacteria ., A/J and C57BL/6J mice were purchased from the Jackson Laboratory ( Bar Harbor , Maine ) ., Thirty-five independent RC strains originally derived from a reciprocal double backcross between the A/J and C57BL/6J progenitors 34 were purchased from Emerillon Therapeutics Inc . ( Montreal , Qc . ) ., All mice were housed in the rodent facility of the Montreal General Hospital ., Animal use protocols were approved by the Animal Care Committee of McGill University and are in direct accordance with the guidelines outlined by the Canadian Council on Animal Care ., Recombinant BCG Russia ( ATCC 35740 ) and Pasteur ( ATCC 35734 ) , were transformed with pGH1 , an integrating vector that inserts into the attB site of the mycobacterial genome and that combines a firefly luciferase lux gene cassette , an integrase int gene , a MOP promoter , and a hygromycin resistance Hyg gene 31 ., The pGH1 vector allows for growth on antibiotic-containing media to reduce risk of contamination 36 ., BCG strains were grown on a rotating platform at 37°C in Middlebrook 7H9 medium ( Difco Laboratories , Detroit , Mich . ) containing 0 . 05% Tween 80 ( Sigma-Aldrich , St . Louis , Mo . ) and 10% albumin-dextrose-catalase ( ADC ) supplement ( Becton Dickinson and Co . , Sparks , Md . ) ., At an optical density ( OD600 ) of 0 . 4 to 0 . 5 , bacteria were diluted in phosphate buffered saline ( PBS ) to 105 colony forming units ( CFU ) /ml ., Mice were injected intravenously with 103 to 104 CFU of BCG in 100 µL of PBS ., Inoculum doses were confirmed by plating on Middlebrook 7H10 agar ( Difco Laboratories , Detroit , Mich . ) supplemented with oleic acid-albumin-dextrose-catalase ( OADC ) enrichment ( Becton Dickinson and Co . , Sparks , Md . ) ., Infected mice were sacrificed by CO2 inhalation after 1 and 6 weeks post-infection ., Lungs and spleens were aseptically removed , placed in 0 . 025% Saponin-PBS , and homogenized mechanically using a Polytron PT 2100 homogenizer ( Brinkman Instruments , Westbury , NY ) ., Homogenates were serially diluted tenfold and plated on Middlebrook 7H10 agar supplemented with OADC enrichment and containing hygromycin B ( Wisent Inc . , St . -Bruno , Qc . ) ., Bacterial enumeration was performed following a six-week incubation at 37°C ., For BCG Pasteur infection , a total of 221 and 175 mice were used at the week 1 and 6 time points , respectively ., A total of 145 and 189 mice , respectively , were used at 1 and 6 weeks for BCG Russia infection ., Strains of the AcB/BcA panel were genotyped for 625 microsatellite markers spanning the entire genome with an average distance of 2 . 6 cM 34 ., Based on Build 36 . 1 of Mouse Genome Informatics ( MGI ) Mouse Genome Database , six markers with reassigned positions were removed from the current analysis 37 ., The first QTL model was the linear model where y represents a vector with the individual total count of bacteria ( log10CFU ) ; is a vector with each entry being an indicator variable of the genotype BB at the marker position with being its associated effect ( major gene effect ) ; is a matrix of fixed covariates ( a constant and gender in our main model ) and its corresponding parameter vector ; is a vector of independent and identically distributed random variables representing the error term with and ., At each marker position , M-estimates of the parameters and a t-statistic were computed ., The genome-wide corrected p-values were obtained by bootstrap under the hypothesis that there is no major gene , i . e . , re-sampling under the reduced model Mean confidence bounds at each marker were defined as twice the standard error around the markers group mean without considering gender effect in the model ., In order to account for the genetic background , a second linear model of the form was employed , i . e . , our second model was the mixed model resulting from adding a random component , to our original model , where is a random vector associated to the genetic background of each RCS and is the design matrix associating the RCS effect to the phenotype y ., The assumptions for this model component were and with being an unknown constant and a positive definite-matrix ( in fact , a background correlation matrix which is a function of length of the segments identical by descent shared amongst strains ) assumed to be known , although a genomic estimate of it was previously obtained ., At each marker position iteratively , estimates of fixed effect parameters and the variance components were obtained under this model and a t-statistic of the same form as before was computed ., The genome-wide corrected p-values were obtained by bootstrap under the hypothesis that there is no major gene , i . e . , re-sampling under the reduced model More details of estimation and testing are given in Methods S1 ., Evidence was considered significant for linkage when single-point regression analysis at the markers was P<0 . 01 ., We determined the bacillary load of BCG strains Pasteur and Russia in the lungs and spleens of C57BL/6J and A/J mice following a low dose ( ∼3×103 bacilli ) intravenous injection of bacilli ., Pulmonary counts of BCG Pasteur were below the limit of detectability ( 80 bacilli/lung ) at weeks 1 and 6 post infection but showed a modest peak of approximately 100 bacilli/lung at week 3 ( Figure 1 ) ., This suggested limited dispersion and growth of BCG Pasteur in the lungs ., In addition , there was no detectable difference in the pulmonary load of BCG Pasteur between C57BL6/J and A/J mice ., By contrast , we observed an increase of 1–1 . 5 log CFU in the spleens between weeks 1 and 3 post infection that was followed by a 1 log decrease at week 6 ., The splenic bacillary burden of BCG Pasteur was substantially higher in C57BL/6J mice at weeks 1 and 3 ., BCG Russia showed a constant increase of pulmonary CFU from week 1 to week 6 ., In the spleen , growth of BCG Russia lagged growth of Pasteur and did not show evidence for a peak at 3 weeks post infection , as was observed for Pasteur ( Figure 1 ) ., Overall , the pattern of tissue CFU for BCG Pasteur strongly resembled the one described for BCG Montreal which has previously been shown to be under Nramp1 control 35 , 38 ., The kinetics of lung and spleen bacillary counts of BCG Russia were distinct from the previously described BCG growth patterns ., To investigate the genetic control of in-vivo growth of BCG Russia and BCG Pasteur , mice from a panel of 35 AcB/BcA RC strains were intravenously challenged with a low dose ( 3–5×103 bacilli ) of BCG Russia or BCG Pasteur ., The number of colony forming units ( CFU ) in the spleen and lung was used as the phenotype for the genetic analysis ., CFU were determined at 1 week and 6 weeks post infection since it is well established that at 3 weeks , the Nramp1 gene dominates the host response to BCG Montreal 38 , making it potentially more difficult to discern additional genetic control elements ., To best indicate the effect of genotype on CFU , all RCS were stratified according to genotype at each marker , i ., e ., AA for markers on chromosomal segments derived from A/J or BB for chromosomal segments derived from C57BL/6J ., Mice of all RCS with a given genotype were then used to obtain the mean and 95% confidence interval of their pulmonary and splenic CFU ., This presentation allowed to graphically depict the effect of both marker genotype and of the general strain background on CFU ., Results for the spleen and lung for both BCG strains are presented in Figures 2 and 3 ., A clear impact of strain background on susceptibility to BCG in the spleen at 1 week post infection was evidenced by the larger bacillary counts in mice of the BB genotype across most chromosomes ( Figure 2 ) ., The strong strain background effect on splenic CFU was resolved by 6 weeks post infection , particularly for BCG Pasteur where differences in splenic bacillary burden appeared negligible across all markers ( Figure 2 ) ., By contrast , CFU differences in BCG Russia were observed for several small chromosomal segments possibly suggesting the presence of specific genetic loci ( Figure 2 ) ., As in the parental strains , pulmonary burdens were at the limit of detectability at week 1 for both Russia and Pasteur , and week 6 for Pasteur ., However , at the 6-week endpoint , preferential replication of BCG Russia was observed in mice bearing specific A/J-derived chromosomal segments , particularly at the distal portion of chromosome 11 ( Figure 3 ) ., Markers where the mean CFU of the AA and BB genotype groups diverged were indicative of chromosomal regions that potentially harboured a BCG susceptibility locus ., To confirm the potential linkage of these chromosomal segments to bacterial burden , a genetic analysis comparing mice of the AA to BB genotype was performed ., The initial analysis compared genotype groups without taking into account the genetic background of the strain or the gender of the mouse ( incomplete model ) ., As expected , markers significantly linked to bacterial burden corresponded well with chromosomal regions where the two genotypes differed ( Figures 2 and 3; Figures S1 to S3 ) ., From this analysis , the genetic control of BCG Pasteur and Russia splenic infection appeared to be highly multigenic at the early time point ., Employing a very stringent level of significance ( P<0 . 0003 ) , quantitative trait loci ( QTL ) were identified across 8 and 15 different chromosomes for BCG Pasteur and Russia , respectively ( Figure S1 ) ., At the 6 week endpoint , a locus was identified on chromosome 1 for splenic BCG Russia load whereas genetic effects were not detected for BCG Pasteur load ( Figure S2 ) ., Pulmonary CFU of BCG Russia was controlled by a locus on chromosome 11 while for BCG Pasteur a locus was identified on chromosome 8 ( Figure S3 ) ., Visual inspection of CFU across genotypes suggested a strong impact of strain background on bacillary loads ., To account for the potential impact of background genes on linkage peaks , we developed a main model that accounted for the genetic background and gender of the mice ., The number of loci identified by the main model was reduced relative to the incomplete model , particularly at the 1 week time point ( Figures S4 and S5 , and Table 1 ) ., For lung CFU , the locus on chromosome 11 remained that impacted on bacillary load of BCG Russia at 6 weeks post infection ( Figure 4 ) ., No genetic effect was detected for pulmonary load of BCG Pasteur which is consistent with the very limited growth of BCG Pasteur in the lungs of all mice ( data not shown ) ., In contrast to the lung , the genetic control of splenic bacillary load remained largely multigenic even after correction for strain background effects ., For BCG Russia at 1 week post infection , a single locus on chromosome 1 ( 36 . 9 cM–48 . 8 cM ) was found to control splenic load ( Figure S4 ) ., At 6 weeks post infection , the genetic control of BCG Russia was multigenic ( Figure S5 ) ., In addition to the chromosome 1 locus ( 32 . 8–55 . 1 cM ) , loci were detected on chromosome 6 ( 45 . 5–46 . 3 cM ) , chromosome 11 ( 47 . 67 cM ) and chromosome 19 ( 51 cM ) ., Splenic load of BCG Pasteur at 1 week post infection was controlled by loci on chromosome 2 ( 10–15 and 22 . 5–26 . 2 cM ) , chromosome 7 ( 63 . 5–65 . 6 cM ) and the X chromosome ( 37–40 . 2 cM ) ., Additional weaker effects were identified on chromosome 3 ( 33 . 7 and 58 . 8 cM ) , chromosome 6 ( 63 . 9 cM ) , chromosome 10 ( 3 cM ) , and chromosome 17 ( 23 . 2 cM ) ., A major gene effect detected on chromosome 1 ( 17–58 . 5 cM ) overlapped the chromosome 1 locus controlling BCG Russia infection ( Figure S4 , Table 1 ) ., Genetic control elements were not detected in response to BCG Pasteur infection at the 6 week time point ( data not shown ) ., The inverse complexity of BCG Pasteur ( multigenic at 1 week; no genes at week 6 ) and BCG Russia ( a single gene at week 1 , multigenic at week 6 ) reflects differences in the replication pattern of the bacteria: BCG Russia showed a delayed onset of growth that continued at week 6 while BCG Pasteur showed rapid initial growth with a strong decline of CFU at week 6 as compared to week 3 ., The chromosome 1 locus significant for linkage early during BCG Pasteur infection and at the early and late phase of BCG Russia infection was indistinguishable from Nramp1 ., Employing what we termed the “conditional model , ” we determined whether the additional linkage peaks were conditional on the Nramp1 gene ., For this the main model was modified to adjust for the effect of Nramp1 by adding a column with the BB genotype indicator at the Nramp1 position to the matrix X . Chromosomal regions identified at the week 1 time point of both BCG Pasteur and BCG Russia infection were no longer significant for linkage following correction for the chromosome 1 locus ( data not shown ) ., Similarly , the genetic effects detected on chromosome 6 and 19 were no longer significant at the 6 week time point of BCG Russia infection ., However , the linkage hit detected on chromosome 11 ( 47 . 67 cM ) retained its significance ., By contrast , a secondary peak detected only for splenic CFU immediately proximal to this locus did not reach significance ( Figure 5 ) ., Finally , an additional locus was localized to chromosome 13 ( 73–75 cM ) ( Figure 5 ) ., RC strains are particularly useful to establish pathways of causality in complex read-outs such as immune reactivity and are well suited to track gene-gene interactions 33 ., However , RC strains have also proven useful for positional identification of disease susceptibility loci by employing RC strains with extreme phenotypes in subsequent genetic crosses 39 , 40 , 41 ., A third application of RC strains is the genome-wide identification of quantitative trait loci ( QTL ) in complex diseases ., This feature of RC stains is particularly attractive since it allows the measurement of quantitative traits in many genetically identical mice belonging to the same strain which greatly increases the accuracy of trait determination ., A genome-wide scan for the presence of QTL can then be conducted among the relatively limited number of RC strains in each panel ., This is highly efficient compared to the breeding and genotyping of hundreds of mice in traditional backcross or F2 based genome-wide mapping studies ., For example , a recent study used the AcB/BcA RC strain panel to localize a large number of asthma susceptibility loci across the genome 42 ., A potential problem that is faced in these speedy genome-wide scans in RC strains is the confounding impact of strain background and of strong susceptibility loci on the overall pattern of QTLs mapped ., We have developed a new analytical methodology that overcomes both of these potentially confounding limitations while conducting genome-wide QTL mapping in RC strains ., Our results demonstrate the ease of genome-wide scanning in RC strains and the importance of adjusting especially on strain background to achieve reliable QTL identification ., Our ability to detect the Nramp1 genomic region also served as an internal validation of the analytical approach ., Another interesting observation was the loci that could only be detected in connection with Nramp1 ., Once the analysis was adjusted on the Nramp1 gene , these loci were no longer significant for linkage ., The most parsimonious explanation for this effect is that these loci are interacting with Nramp1 ., Why we would detect a large number of genes that interact with Nramp1 in the genetic control of BCG Pasteur as compared to BCG Russia is not known but may reflect the differences in pathogenesis between the two BCG strains ., For BCG Pasteur , putatively interacting genes were detected at 3 weeks post infection while for BCG Russia such interacting loci were observed at the 6 week time point ., At 3 weeks , BCG Pasteur shows a sharp peak of splenic bacillary burden while the growth of BCG Russia continues well past 6 weeks before a slow and gradual reduction of splenic burden becomes evident after 12 weeks of infection ( data not shown ) ., While the interpretation of our results as Nramp1 interacting loci appears reasonable , it is important to realize that this conclusion needs further direct experimental validation ., However , if correct , the mapping tools presented in this paper would provide a very powerful approach for the identification of interacting loci which is still a major obstacle in complex trait analysis in both human and model animals ., The study of the impact of strain variability of M tuberculosis on disease expression is of considerable interest for the implementation of tuberculosis control measures ., An increasing body of evidence suggests that different strains/lineages of M . tuberculosis display substantial differences in their pathogenic potential 8 , 9 ., In addition , evidence is emerging that genetic variability among BCG vaccine strains is a potent factor in modulating BCG induced anti-tuberculosis immunity 31 ., This mycobacterial strain variability reflects an even greater divergence in host responsiveness to both BCG and M . tuberculosis that is largely under host genetic control ( reviewed in 43 ) ., These observations raise the question if host and mycobacterial variability are independent of each other ., If independent , we would expect hosts to display a spectrum of responsiveness from highly resistant to highly susceptible irrespective of the infecting mycobacterial strain ., Similarly , M . tuberculosis strains would vary from highly virulent to mildly virulent across all hosts ., Alternatively , it is possible that “susceptibility” and “virulence” are not absolute but rather reflect specific combinations of mycobacterial strain and human host ., The latter possibility is supported by recent observations of preferential associations of tuberculosis lineages with ethnic groups that may reflect co-adaptation of M . tuberculosis and its human host 10 ., Moreover , a number of host genetic association studies have reported a preferential association of tuberculosis susceptibility variants with specific M . tuberculosis lineages 12 , 13 , 14 ., The results of our study obtained in a highly controlled experimental setting support the hypothesis of host – pathogen specific genetic “fits . ”, Hence , human susceptibility to tuberculosis may only become tractable by jointly considering host and pathogen genetic backgrounds ., By conducting a genome-wide mapping of loci that impact on the splenic and pulmonary burden following a low dose infection with two strains of BCG , we revealed a divergent pattern of susceptibility loci ., An unexpected result was the pronounced dynamic of genetic loci impacting on bacillary counts ., This observation demonstrated how different genetic control elements came into play as the BCG infection advanced and further emphasized the intimate interplay between host genetics and pathogenesis ., Perhaps less surprising was the large difference in the number of loci involved in the control of splenic vs pulmonary bacillary counts ., BCG Pasteur shows little dissemination and growth in the lungs of infected mice and the absence of susceptibility loci was therefore expected ., However , BCG Russia reaches bacillary counts in the lungs that are similar to those in the spleen ., Yet , only one susceptibility locus on chromosome 11 was detected to impact on pulmonary counts while splenic counts are under more complex control ., It is interesting that a locus on chromosome 1 which is indistinguishable from the Nramp1 gene had by far the strongest impact on bacillary burden in both BCG Pasteur and Russia , but this effect was limited to splenic counts ., By contrast , the chromosome 11 locus was detected only for BCG Russia but in both the spleens and lungs ., The results therefore indicate that host genetic control is characterized by very strong common control elements that act in a tissue –specific manner , and by somewhat weaker BCG strain specific susceptibility genes that are not tissue specific ., Together these data indicate that host genetic control of mycobacterial replication is sensitive to the particular strains but also to differences in disease manifestations ( here , lung vs spleen ) ., Interestingly , the strongest genetic effect ever found in human studies was found in an outbreak of tuberculosis in Northern Canada 44 ., During this outbreak , all cases had been infected from a single index case , i . e . a single bacterial strain 45 ., A fine tuned host genetic response to mycobacteria might explain why it has been difficult to reproducibly detect strong host genetic effects in human tuberculosis ., Consequently , future genetic studies of tuberculosis susceptibility might need to be adjusted on the detailed clinical picture and infecting M . tuberculosis strain . | Introduction, Materials and Methods, Results, Discussion | The host response to mycobacterial infection depends on host and pathogen genetic factors ., Recent studies in human populations suggest a strain specific genetic control of tuberculosis ., To test for mycobacterial-strain specific genetic control of susceptibility to infection under highly controlled experimental conditions , we performed a comparative genetic analysis using the A/J- and C57BL/6J-derived recombinant congenic ( RC ) mouse panel infected with the Russia and Pasteur strains of Mycobacterium bovis Bacille Calmette Guérin ( BCG ) ., Bacillary counts in the lung and spleen at weeks 1 and 6 post infection were used as a measure of susceptibility ., By performing genome-wide linkage analyses of loci that impact on tissue-specific bacillary burden , we were able to show the importance of correcting for strain background effects in the RC panel ., When linkage analysis was adjusted on strain background , we detected a single locus on chromosome 11 that impacted on pulmonary counts of BCG Russia but not Pasteur ., The same locus also controlled the splenic counts of BCG Russia but not Pasteur ., By contrast , a locus on chromosome 1 which was indistinguishable from Nramp1 impacted on splenic bacillary counts of both BCG Russia and Pasteur ., Additionally , dependent upon BCG strain , tissue and time post infection , we detected 9 distinct loci associated with bacillary counts ., Hence , the ensemble of genetic loci impacting on BCG infection revealed a highly dynamic picture of genetic control that reflected both the course of infection and the infecting strain ., This high degree of adaptation of host genetics to strain-specific pathogenesis is expected to provide a suitable framework for the selection of specific host-mycobacteria combinations during co-evolution of mycobacteria with humans . | Susceptibility to mycobacterial infection results from a complex interaction between host and bacterial genetic factors ., To examine the effect of host and pathogen genetic variability on the control of mycobacterial infection , we infected a panel of genetically related recombinant congenic ( RC ) mouse strains with two closely related strains of Mycobacterium bovis BCG ., Bacterial counts of BCG Russia and BCG Pasteur were determined in the lung and spleen at 1 and 6 weeks following infection and used for genetic analysis ., A novel analytical approach was developed to perform genome-wide linkage analyses using the RC strains ., Comparative linkage analysis using this model identified a strong genetic effect on chromosome 1 controlling counts of BCG Pasteur at 1 week and of BCG Russia at 1 week and 6 weeks in the spleen ., A locus impacting on late BCG Russia counts in the lung and spleen was identified on chromosome 11 ., Nine additional loci were shown to control bacterial counts in a tissue- , time- , and BCG strain-specific manner ., Our findings suggest that the host genetic control of mycobacterial infection is highly dynamic and adapted to the stage of pathogenesis and to the infecting strain ., Such a high degree of genetic plasticity in the host-pathogen interplay is expected to favour evolutionary co-adaptation in mycobacterial disease . | infectious diseases/bacterial infections, genetics and genomics/complex traits, infectious diseases, genetics and genomics | null |
journal.ppat.1004758 | 2,015 | Spatiotemporal Analysis of Hepatitis C Virus Infection | Hepatitis C virus ( HCV ) belongs to the Flaviviridae family of enveloped , positive-stranded RNA viruses ., Following productive entry into hepatocytes , the 9 . 6 kb HCV genome is translated to produce a single large polyprotein 1 , which is cleaved by viral and host proteases to yield ten distinct protein products 2 ., These proteins include the structural proteins ( core , E1 and E2 ) and the non-structural proteins ( p7 , NS2 , NS3 , NS4A , NS4B , NS5A , and NS5B ) ., The five “replicase” proteins NS3 to NS5B are essential and sufficient for HCV RNA replication 3 , 4 ., Similar to all other positive strand RNA viruses , HCV induces rearrangements of intracellular membranes to create a favorable microenvironment for RNA replication to occur 5–8 ., Replication complex formation appears to require the viral NS4B and NS5A proteins 5 , 9 ., NS5B , the viral RNA-dependent RNA polymerase is the key enzyme of the replicase complex 10 , 11 ., Using the ( + ) strand genome as a template , NS5B first synthesizes a complementary ( − ) strand , resulting in a double-stranded ( ds ) RNA intermediate , and then proceeds to transcribing progeny ( + ) strands ., Newly synthesized ( + ) strand RNAs are then thought to be shuttled out of replication compartments to serve as templates for further translation by cellular ribosomes or become encapsidated into assembling virions on the surface of lipid droplets ( LDs ) 12 ., Although these processes are likely linked , a single viral ( + ) strand RNA can only be involved in either translation , replication or packaging at a given time , and the switch from one process to another has to be regulated 13 ., For HCV , the switch from translation to replication is unclear ., The cellular protein Ewing sarcoma breakpoint region 1 ( EWSR1 ) binds to the viral RNA cis acting replication element ( CRE ) , and has been proposed to regulate the switch from translation to replication by modulating the kissing interaction between the CRE and a RNA stem-loop structure in the HCV 3’ UTR 14 ., Similarly , for polioviruses , the switch from translation to replication is regulated by the action of viral proteases on a cellular protein binding to the 5’cloverleaf viral RNA structure 15 ., The switch from replication to assembly is not well understood , however it has been suggested that the phosphorylation state of NS5A might regulate the process 16 ., It is also possible that HCV ( + ) RNA fate is spatially regulated by the distinct subcellular localizations of translation , replication and assembly ., It is not clear how HCV spatially and temporally regulate its lifecycle within the host cell ., Mathematical models have been developed to study HCV RNA dynamics during primary infection of chimpanzees 17 , and humans ( liver transplantation 18 , 19 , response to IFN 20 and ribavirin treatment 21 ) , in addition to HCV replicons in cell culture 22–25 ., Recently , data on the dynamics of infectious HCV RNA in cell culture became available 26 ., While these past studies were focused on whole cell populations , organs and organisms , data on viral RNA kinetics in single cells are lacking ., Advances in single RNA detection methods using ISH followed by amplification 27 enable the analysis of HCV RNA at the single cell level ., Previous studies have utilized this approach to detect ( + ) strand HCV RNAs in hepatoma cell lines 28 as well as in infected human liver biopsies 29 ., To better understand the spatiotemporal organization of HCV infection , we have developed RNA detection methods that allow for simultaneous visualization of HCV ( + ) and ( − ) RNA strands at the single cell level ., We have combined this approach with 4-color high-resolution confocal microscopy to study the localization of ( + ) and ( − ) HCV RNAs with viral and host proteins of interest ., Using this single cell RNA detection approach , we can interrogate the cell-to-cell variability of HCV infection kinetics ., Moreover , we can determine whether ( + ) strand fates are regulated temporally ., The approach developed in this study can be generically applicable to all RNA viruses and enables unprecedented sensitivity for studying early events in the viral life cycle , which are unappreciated for viruses of relatively low replicative capacity , including some viruses of the Flaviviridae , due to limits of detection ., We used the QuantiGene ViewRNA ISH detection system to specifically detect HCV ( + ) and ( − ) strand RNAs in single cells ., Briefly , non-overlapping oligonucleotide probe sets specific to either the ( + ) or ( − ) strand of HCV NS3/4A region were hybridized to target RNAs and sequential hybridization steps provided up to 8 , 000-fold signal amplification ., To validate the specificity of labeling , wild type or polymerase defective ( GND ) subgenomic JFH1 HCV replicon RNAs were electroporated into Huh-7 . 5 cells , which were then fixed and processed for HCV RNA detection ( Fig . 1A ) and quantification ( Fig . 1B ) ., ( + ) strand RNA , comprising a mix of input and newly synthesized RNA , was abundant at 6 hours post-electroporation ( hpe ) and increased up to 20 fold by 96 hpe for the wild type , but not the polymerase defective replicon containing cells ., ( − ) strand RNA was also detected at 6 hpe and increased at a similar rate as the ( + ) strand by 96 hpe for wild type HCV ., There was no ( − ) strand detected at any of the time points for the GND replicon , thus confirming that polymerase function is required for ( − ) strand RNA detection ., As expected , there was no signal for either ( + ) or ( − ) HCV RNAs detected in mock infected cells ., To further confirm the specificity of our ( − ) strand RNA labeling in the context of HCV infection , we used the HCV direct-acting antivirals Sofosbuvir and Daclatasvir ., Sofosbuvir is a nucleotide analogue inhibitor of the NS5B polymerase 30 already approved for anti-HCV therapy in combination with ribavirin or other direct acting antivirals such as Daclatasvir , Ledispavir , or Simeprevir 31 ., Daclatasvir is a potent inhibitor of the NS5A protein 32 that is already approved for anti-HCV therapy in Europe and Japan 33 ., As expected , there was very little ( − ) strand accumulation at 6 and 48 hours post-infection ( hpi ) in cells treated with either Sofosbuvir or Daclatasvir at time of infection ., In contrast , ( − ) strand was detected at 6 hpi and increased 10-fold by 48 hpi in DMSO treated cells ( S1 Fig . ) ., We did not observe an effect of either inhibitor on ( + ) strand levels at 6 hpi , however by 48 hpi there was a drastic decrease in ( + ) strand numbers due to the replication defect imposed by the antiviral drugs ., The kinetics of accumulation and ratios of HCV ( + ) and ( − ) strand RNAs are unknown at the single cell level ., To this end , we quantified HCV ( + ) and ( − ) strand RNAs at the single cell level over a time course of infection ., As expected , we observed ( + ) and ( − ) strand HCV RNA numbers increased over time ( Fig . 2A ) ., Quantitation of ( + ) and ( − ) strand RNAs showed considerable cell-to-cell variability ( >10-fold ) with respect to the number of RNA puncta per cell plane ( Fig . 2B ) ., ( + ) strand puncta could be detected as early as 2 hpi in ~80% of cells , which likely reflected input genomic RNA ., ( − ) strand puncta were reliably detected in cells at 6 hpi , suggesting that viral RNA replication is established by 6 hpi ., Consistent with this interpretation , we observe modest increases in ( + ) RNA at 6 and 12 hpi ., Much larger increases in ( + ) and ( − ) RNA accumulation occurred between 12 and 24 hpi , suggesting that this time point correlates with more robust HCV replication ., By 48 hpi , an average of 331±52 ( + ) RNAs and 49±12 ( − ) RNAs were observed per cell plane ., In order to correlate this value with HCV RNAs per cell , we performed Z-stack analysis of HCV infected cells at 48 hpi and found that the average per cell number was 415±39 ( + ) RNAs and 94±11 ( − ) RNAs per cells ( S1 Movie ) ., Thus , the fluorescence per cell plane captures the majority of the whole cell RNAs ., This high percentage reflects the high fluorescence of the RNA ISH signal , in addition to the relative flatness of Huh-7 . 5 cells ., We observed a ( + ) strand to ( − ) strand ratio of about 10:1 throughout most of the time course , which is in agreement with the ratio determined in infected hepatocytes in humans 34 as well as previous reports using subgenomic replicon systems 3 , 35 ., Notably , this ratio was about 6:1 early in infection ( 6 hpi ) , which correlates with the ratio of ( + ) : ( − ) strands inside replication complexes 35 ., Although the average ( + ) : ( − ) RNA ratio of ~10 was relatively constant , there was significant cell-to-cell variability in ratio , ranging from 1–35 ., We next developed a series of microscopy assays to visualize the association of HCV RNA with markers of specific stages of the viral life cycle ., HCV translation was defined as the colocalization of ( + ) RNAs with actively translating ribosomes ( Fig . 3 ) ., Active RNA replication was defined as colocalization of HCV ( − ) RNA with ( + ) RNA or replicase components NS3 and NS5A ( Fig . 4 , Fig . 5 ) ., HCV assembly was defined as colocalization of ( + ) RNA with core and intracellular virions were defined as colocalization of ( + ) RNA with virion E2 ( Fig . 6 ) ., A caveat to the interpretation of these assays is that colocalization of HCV RNA with these markers does not define a physical interaction ., ( + ) RNA viruses cannot be simultaneously translated ( with ribosomes moving 5’ to 3’ ) and replicated ( with the replicase moving 3’ to 5’ ) 13 ., These processes need to be coordinately regulated , either by spatially separating ( + ) strand RNAs destined for translation and replication , or by regulating the initiation of these processes ., In the case of HCV , the mode of regulation is not yet known , nor is it known whether translation occurs in the proximity of replication compartments ., To address these questions , we applied a ribosome-bound nascent chain puromycylation assay 36 that detects actively translating ribosomes in our single molecule HCV RNA detection assay ., Briefly , puromycin , a Tyr-tRNA mimetic , enters the ribosome A site and terminates translation ., Puromycylated nascent chains are stalled on ribosomes by cycloheximide , a chain elongation inhibitor , and detected via fixed cell immunofluorescence using an anti-puromycin monoclonal antibody ., In the absence of puromycin , we detect no fluorescent signal , while puromycin labeling is readily detected in the presence of puromycin ( Fig . 3A ) , primarily in the cytoplasm but also some in the nucleus , consistent with a previous report 36 ., To confirm the specificity of puromycin labeling , cells were pre-treated with anisomycin , a competitive inhibitor of translation , prior to puromycylation ., As expected , anisomycin drastically decreased puromycin labeling ( Fig . 3A ) ., We next performed the puromycylation assay in combination with RNA ISH over a time course of HCV infection ( Fig . 3B , C ) ., ( + ) RNA translation , as defined by ( + ) RNA localization with puromycylated ribosomes , occurred as early as 2 hpi , with a peak in the percent of ( + ) RNAs undergoing translation ( 70% ) at 6 hpi ., Between 12 and 24 hpi , a steady state level of translated ( + ) RNAs was achieved at ~35% ., While the ( − ) RNA is unlikely to be associated with actively translating ribosomes , we do observe that a high percentage of HCV ( − ) RNAs also colocalize in the proximity of puromycylated ribosomes ., This indicates that sites of viral RNA replication and translation may be in close proximity ., Interestingly , we observed an unusual puromycylation staining pattern in ~20% of the cells at 6 hpi , in which enlarged puromycylation puncta were detected ( Fig . 3B ) ., It is possible that these enlarged puncta reflect ER rearrangements ( Fig . 3B ) ., In support of this interpretation , we observed a similar localization pattern of the ER marker calnexin at this time point of infection ( Fig . 3D ) ., HCV RNA synthesis involves a dsRNA intermediate , which localizes inside the viral induced replication complexes 9 , 37 ., We next quantified ( + ) and ( − ) RNA colocalization over a time course of infection as a marker of HCV RNA replication ( Fig . 4A , B ) ., We observed that ( − ) strand puncta colocalizing with ( + ) strand puncta became detectable as early as 4 hpi; however , it was generally a low frequency event from 4–12 hpi ., The frequency of HCV ( + ) and ( − ) RNA colocalization increased significantly between 12 and 24 hpi , with ~25–35% of ( − ) strand puncta colocalized with ( + ) strand puncta ., This percentage remained relatively constant from 24–72 hpi , suggesting that only ~1/3 of ( − ) strands are actively engaged in RNA replication at a given time ., Unfortunately , attempts to correlate the ( + ) and ( − ) RNA colocalization with dsRNA localization using the commonly used J2 dsRNA antibody were unsuccessful , due to incompatible fixation and permeabilization conditions ., We next examined the localization of HCV RNAs with viral protein components of the replication complex ., HCV NS5A is a multifunctional protein involved in both the replication and assembly stages of HCV life cycle 4 , 38–40 ., NS5A appears to have at least two functions in HCV RNA replication ., It is part of the replicase complex that binds viral RNA 3 , 4 and it also promotes the formation of double membrane vesicles , which are thought to be the sites of viral RNA replication 9 ., The function of NS5A in replication complex formation is regulated in part by the interaction of NS5A with its host cofactor , phosphatidylinositol-4-kinase III-α ( PI4KA ) 41–43 ., Additionally , NS5A partially colocalizes with core protein on surface of lipid droplets and is required for virion assembly 12 , 38 ., We performed a time course of ( + ) and ( − ) strand RNA colocalization with the NS5A protein during HCV infection ( Fig . 5A ) ., Since NS5A protein levels early in infection are undetectable by standard immunofluorescence analysis we used a tyramide signal amplification system ( TSA ) to detect NS5A at 6 and 12 hpi ., We observed limited colocalization of NS5A with either ( + ) or ( − ) RNA at 6 hpi; however , there was a dramatic , specific increase in NS5A colocalization with ( − ) RNA at 12 hpi ( ~90% of ( − ) RNA colocalized with NS5A , Fig . 5B ) ., At later time points , ~30% of NS5A colocalized with ( − ) RNA , which would suggest , only ~30% of ( − ) strand RNAs undergo replication at later times of infection , which is consistent with the observed colocalization of ( − ) with ( + ) RNA ., The localization of NS5A with ( + ) RNA increased over time , consistent with its role in virion assembly ( Fig . 5B ) ., NS3 is also part of the replicase complex and its helicase activity is required for HCV replication and possibly assembly 7 , 44 , 45 ., In contrast to NS5A , NS3 did not preferentially colocalize with ( − ) strands at early time points ( Fig . 5C , D ) ., At later time points , we observed that NS3 had similar levels of RNA colocalization as was observed for NS5A ( ~30% ) ., The distinct frequencies of ( − ) RNA localization with NS3 and NS5A are consistent with the interpretation that NS5A , but not NS3 , is involved in the initial formation of replication compartments , while both NS3 and NS5A are involved in replicase function ., The current model for HCV assembly is that ER-derived HCV replication complexes are in close proximity to the sites of virion assembly , intracellular lipid droplets ( LDs ) ., The viral core ( capsid ) protein accumulates on or near the surface of LDs and mutants that attenuate virion assembly lead to a hyper-accumulation of core at the LD 6 , 46 , 47 ., Viral RNA containing capsids are then enveloped at the ER and egress cells via the secretory pathway in association with components of the VLDL machinery 48–50 ., We first evaluated the compatibility of the RNA ISH protocol with detection of virion associated HCV RNA ., Buoyant density purified HCV virions were processed for RNA detection followed by immunostaining for core protein ., As shown in S2 Fig ., , ~60% ( + ) RNA puncta ( red ) colocalize with core puncta ( green ) ., Thus , single molecule virion RNAs can be detected by the RNA ISH protocol ., We then quantified ( + ) and ( − ) strand RNA colocalization with HCV core at 24 , 48 and 72 hpi ( Fig . 6A , B ) ., As expected , we observed ( + ) strand colocalization with core that increases with time ., ~15% of ( + ) RNA is colocalized with core at 24 hpi , which increases to ~35% at 72 hpi ., We did observe some ( − ) strand colocalization with core; however it did not increase over the time course , suggesting that it was unrelated to virion assembly ., An intensity line profile ( Fig . 6C ) showed that the ( − ) strand ( magenta ) signal , although “colocalized” does not identically overlap with LD-associated core , but is instead juxtaposed to the LD-associated core ., In contrast , the ( + ) strand ( red peak ) overlaps the core ( green ) peak suggesting full colocalization ., Thus , residual localization of ( − ) RNA with core likely reflects the close proximity of sites of replication and assembly ., To quantify intracellular assembled virions we made use of an antibody ( CBH-5 ) that recognizes E2 on fully assembled virions 51 ., This E2 localization pattern is very different from the pattern of E2 localization using E2 antibodies that recognize ER localized E2 ( S3 Fig . ) ., Furthermore , CBH-5 staining in an established HCV assembly mutant ( NS2-G10P ) 52 is at background levels similar to an uninfected control ( S4 Fig . ) ., At 48 and 72 hpi , approximately 10% of ( + ) strand puncta colocalize with E2 ( Fig . 6D , E ) ., We observe that virtually all of the E2 puncta ( green ) colocalize with ( + ) strand RNAs ( red ) , while minimal colocalization of E2 was observed with ( − ) strand RNAs ( magenta ) ., We next plotted the kinetics of the HCV life cycle based on quantitation of the localization of HCV RNAs with markers of translation ( ( + ) strands with puromycylated ribosomes ) , replication compartment formation ( ( − ) RNA localized with NS5A ) , active replication ( localization of ( − ) RNA with ( + ) RNA and NS3 ) , and assembly and intracellular virions ( ( + ) RNA colocalization with core and virion E2 , respectively ) ., The standard quantitation of HCV RNA by quantitative RT-PCR ( Fig . 7A ) and infectious HCV ( Fig . 7B ) show a typical increase of both over time ., HCV RNA increased at 12–24 hpi and plateaued between 48 and 72 hpi , while intracellular virions peaked between 24 and 48 hpi and plateaued by 72 hpi ., We first quantified the total ( + ) RNAs devoted to translation , replication , assembly , or virions as determined in our previous assays ( Fig . 7C ) ., We observed similar kinetics of ( + ) RNA accumulation as compared to the qRT-PCR assay ., HCV ( + ) RNAs were initially associated with translation , while ( + ) RNAs associated with replication increased from 12 to 48 hpi and then slightly decreased ., HCV RNAs associated with virion assembly increased between 24 and 48 hpi and plateaued ., We then quantified the relative amounts of ( + ) RNA devoted to each process ( Fig . 7D ) ., ( + ) strand RNAs display a defined temporal kinetics , with the majority of ( + ) RNAs associated with actively translating ribosomes at early times of infection , followed by a peak of replication between 12 and 24 hpi , followed by virion assembly and detection of assembled virions in the cell cytoplasm ., Interestingly , after the peaks of ( + ) RNA association with translation and then replication were displaced , both populations achieved a steady state level of ~25% of total ( + ) RNAs each , which slowly decayed over time as the levels of virion-associated ( + ) RNAs increased ., In comparing the relative kinetics of each process in the viral life cycle , the slowest kinetic delay is from HCV assembly to accumulation of intracellular virions ( Fig . 7D ) ., Quantitation of HCV ( − ) RNA colocalization with markers of replication revealed distinct phases of HCV replication ( Fig . 7E ) ., Transient active replication could be observed at early time points ( 4 hpi ) , however , robust active replication did not occur until between 12 and 24 hpi ., This was preceded by a high level of colocalization of NS5A with ( − ) RNA ., These data are consistent with a model wherein low levels of HCV RNA replication precede replication compartment ( or membranous web ) formation , replication compartments are then formed at 12 hpi , which is then followed by robust HCV RNA replication ., The percent of HCV ( − ) RNAs engaged in active RNA replication then remains relatively constant throughout the time course of infection at ~35% ., Recent improvements in multiplexed single molecule RNA FISH enable sensitive single cell detection of viral RNAs in infected cells ., This approach has been used to detect RNA from a number of viruses , including the ( + ) strand RNA of HCV in vitro and in vivo 29 , 53 ., In this study , we developed assays for the specific co-detection of HCV ( + ) and ( − ) RNA in the same infected cell , in combination with high-resolution microscopy to quantify the kinetics of HCV RNAs at the single cell level ., We verified the specificity of these assays ., ( + ) RNA detection was specific to HCV infected- or transfected-cells ., RNA puncta increased in abundance during infection , which was precluded by HCV inhibitors ., HCV ( − ) RNA detection had similar properties , in addition to being specific for cells transfected with wild type , but not polymerase defective , HCV RNAs ( Fig . 1 and S1 Fig . ) ., We confirmed that this approach has single molecule sensitivity and that the fixation/detection method is compatible with detection of HCV RNAs in infectious virions ( S2 Fig . ) ., We then used this assay to perform a kinetic analysis of HCV infection at the single cell level ( Fig . 2 ) ., HCV ( + ) RNA was readily detected at 2 hpi , while the ( − ) RNA was detected at 6 hpi ., Both ( + ) and ( − ) RNA accumulation increased over a 48 hour time course , as would be expected ., There was considerable cell-to-cell variability in both ( + ) and ( − ) RNA accumulation with both showing an ~10-fold range in values at a given time point ., This variability was also reflected in ( + ) / ( − ) RNA ratio , which ranged from less than one to greater than 35 depending on the infected cell ., This large variability in ( + ) / ( − ) ratio suggests that distinct host factors that differentially regulate ( + ) and ( − ) RNA synthesis are limiting in individual Huh-7 . 5 cells ., Despite this cell-to-cell variability , the mean ( + ) / ( − ) RNA ratio of ~10 was relatively constant and is consistent with previous estimates using strand-specific northern blot analysis of HCV replicon cells 3 , 35 as well as qRT-PCR analysis of HCV-infected whole cell lysates 26 ., We next developed microscopic assays to identify HCV RNAs associated with actively translating ribosomes , replication complexes , nucleocapsid assembly , and intra-cellular virions ., The colocalization of HCV ( + ) RNAs with puromycylated ribosomes indicated actively translating HCV RNAs ., At 6 hpi , the majority of ( + ) RNAs were associated with translating ribosomes , and this association subsequently decreased to a steady state of ~30% of total HCV RNAs over the time course of infection ( Fig . 3 ) ., Interestingly , there was also a high level of colocalization of puromycylated ribosomes with HCV ( − ) RNAs ., While it is possible that HCV ( − ) RNAs are associated with ribosomes , a more likely interpretation is that sites of HCV translation and replication are in close proximity ., ( + ) RNAs cannot be simultaneously translated and replicated due to steric hindrance between ribosomes traveling 5’ to 3’ and the HCV replicase moving 3’ to 5’ 13 ., Models for the differential regulation of HCV translation and replication include spatial separation ( e . g . the exclusion of ribosomes from replication compartments ) and/or the differential regulation of these processes via protein-RNA and RNA-RNA interactions ., The viral polymerase NS5B can only initiate RNA synthesis from the same RNA that it has been translated from ( a cis requirement ) 54 ., Additionally , viral translation is dependent on active RNA replication 55 , suggesting that the translation complexes have to be in very close proximity to the replication complexes ., In this case , the decision whether to translate or replicate the viral RNA may be modulated by regulation of HCV RNA-RNA or RNA-protein interactions ., For example , the HCV RNA kissing loop interaction between the HCV cis-acting replication element and the 3’UTR by EWSR1 may promote a switch between HCV translation and replication 14 ., Our data suggest that replication and translation are spatially linked ., Development of puromycylation assays compatible with immuno-EM may answer the question of whether active translation occurs within , or adjacent to , HCV replication compartments ., Detection of the ( − ) RNA , the replication intermediate , is indicative of a viral replication complex ., We detect ( − ) RNA as early as 4 hpi in infected cells , some of which is involved in active replication , as defined by ( + ) RNA colocalization ( Fig . 4 ) ., By 12 hpi the majority of ( − ) RNA is associated with NS5A , but not NS3 or ( + ) RNA ., NS5A has been implicated in the formation of HCV replication compartments 9 , which may explain this selective HCV ( − ) RNA/NS5A colocalization at 12 hpi ., We also observe high levels of colocalization of ( − ) RNA at the 6 hpi time point with large globular fluorescent labeling of ER markers , including ribosomes and calnexin , in a subset of cells ( Fig . 3B , D ) ., Given the kinetic coincidence between ( − ) RNA appearance , NS5A association , and altered ER morphology , it is possible that this represents large scale alterations of the ER associated with replication complex formation ., Alternatively , it may be a stress response to infection , such as induction of the ER stress response ., The appearance of ( − ) RNA and low levels of transient dsRNA replication intermediates at 4 hpi appears to precede HCV replication compartment/membranous web formation , as defined by extensive ( − ) RNA/NS5A colocalization at 12 hpi , and robust HCV RNA replication ( 12–24 hpi ) ., The proposed kinetics of membranous web formation is consistent with previous EM data that described the initial appearance of intracellular double membrane vesicles at 16 hpi and further accumulation until 24 hpi 9 ., The transient HCV replication prior to replication compartment formation suggests a strategy wherein HCV synthesizes limited amounts of ( + ) and ( − ) RNAs early during infection to decrease its reliance in the integrity of the initially infecting HCV ( + ) RNA ., Any damage or modifications to the initial ( + ) RNA would result in an abortive infection , this it makes sense that the virus would synthesize a small pool of viral RNAs early to maximize the probability of a productive infection ., It is unclear whether these early replication intermediates ( 4 hpi ) are shielded by a small amount of membrane remodeling that precedes the robust replication compartment formation at 12 hpi ., If these replication intermediates are not membrane-protected , they may be vulnerable to cytosolic RNA sensors , such RIG-I-like RNA sensors and dsRNA activated protein kinase R ( PKR ) ., It remains to be determined whether these RNAs are recognized by innate immune sensors or alternatively , whether HCV has evolved a mechanism to protect these RNAs from detection ., Future studies will examine the localization of innate immune RNA sensors with HCV ( + ) , ( − ) , and dsRNAs ., Higher proportions of active replication complexes , as defined by colocalization of ( − ) RNA with either ( + ) RNA or NS3 are detected later in infection , between 12 and 24 hpi ., This time point corresponds to increases in ( + ) and ( − ) RNA fluorescent puncta ( Fig . 4 ) , in addition to HCV RNA levels by quantitative real time RT-PCR ( Fig . 7A ) ., The proportion of HCV replication complexes that are defined as active ( % of ( − ) RNAs that colocalize with NS3 or ( + ) RNAs ) is consistently ~30% after 24 hpi ., This is consistent with predictions from cryo-EM analysis of HCV replication complexes , which indicate that some HCV replication compartments are not compatible with active replication complexes 9 ., The specific viral ( − ) strand detection is of practical interest , since despite extensive efforts; there are few good markers of HCV replication complexes for microscopy analysis ., Only a small fraction of the NS5B polymerase ( <5% ) resides inside replication complexes 35 and other replicase proteins ( NS3 and NS5A ) have multiple localizations beyond the replication complex ., Similarly , there are no cellular proteins have been described that are consistently localized solely to replication complexes ., We and many other labs have used a dsRNA antibody to detect putative viral RNA replication complexes 37 , 56 ., Although , this antibody recognizes antigen within replication compartments and is specific to infected cells 9 , it is possible that the antibody may also detect structured viral RNAs ., We attempted to assess the degree of colocalization of the dsRNA antibody with ( − ) RNA; however , the fixation and permeabilization steps of the two detection methods were not compatible ., HCV ( + ) RNAs associated with core both in the proximity of lipid droplets , which are putative sites of nucleocapsid assembly , and in discrete puncta , which may represent assembled virions ., The percent HCV ( + ) RNA that colocalized with core increased during the time course from ~15% at 24 hpi to ~35% at 72 hpi ( Fig . 6 ) ., Similarly the colocalization with ( + ) RNA and an E2 antibody that preferentially recognizes E2 incorporated into virions increased over the time course from virtually none at 24 hpi to ~12% of ( + ) RNAs at 72 hpi ., This kinetics mirrors the production of infectious HCV ( Fig . 7B ) ., We observed a juxtaposition of ~15% of HCV ( − ) RNAs with core , consistent with the spatial linkage between sites of RNA replication and assembly ., No HCV ( − ) RNA colocalized with virion E2 ., The development of assays to define ( + ) RNAs associated with translation , replication , nucleocapsid assembly , and intra-cellular virions allowed us to quantify the total number of ( + ) RNAs , in addition to the proportion of ( + ) RNAs , associated with each process ( Fig . 7 ) ., We observed an initial association of ( + ) RNAs with translation , followed rapidly by replication ., The number of ( + ) RNAs associated with translation or replication increased with similar kinetics and abundance until 48 hpi , after which they diminished slightly ., Association of ( + ) RNAs with nucleocapsid assembly was detected at 24 hpi and peaked at 48 hpi , after which it plateaued as the most abundant class of HCV ( + ) RNAs in infected cells ., Putative association of ( + ) RNAs with intracellular virions was detected at 48 hpi and increased slightly at 72 hpi ., Analysis of the proportion of ( + ) RNAs associated with each process in the viral life cycle revealed a tightly coordinated regulation ., An initial peak of HCV translation at 6 hpi was rapidly displaced by HCV replication at 12–24 hpi , after which both maintained a steady state level of ~25% each of total HCV ( + ) RNAs ., The proportion of ( + ) RNA associated with nucleocapsid assembly began to displace the replication peak at 24 hpi , after which it gradually increased throughout the time course ., Finally , ~10% of HCV ( + ) RNA associated with intracellular virions was detected at 48 hpi and beyond ., The slowest kinetic delay that we observed is from HCV assembly to accu | Introduction, Results, Discussion, Materials and Methods | Hepatitis C virus ( HCV ) entry , translation , replication , and assembly occur with defined kinetics in distinct subcellular compartments ., It is unclear how HCV spatially and temporally regulates these events within the host cell to coordinate its infection ., We have developed a single molecule RNA detection assay that facilitates the simultaneous visualization of HCV ( + ) and ( − ) RNA strands at the single cell level using high-resolution confocal microscopy ., We detect ( + ) strand RNAs as early as 2 hours post-infection and ( − ) strand RNAs as early as 4 hours post-infection ., Single cell levels of ( + ) and ( − ) RNA vary considerably with an average ( + ) : ( − ) RNA ratio of 10 and a range from 1–35 ., We next developed microscopic assays to identify HCV ( + ) and ( − ) RNAs associated with actively translating ribosomes , replication , virion assembly and intracellular virions ., ( + ) RNAs display a defined temporal kinetics , with the majority of ( + ) RNAs associated with actively translating ribosomes at early times of infection , followed by a shift to replication and then virion assembly ., ( − ) RNAs have a strong colocalization with NS5A , but not NS3 , at early time points that correlate with replication compartment formation ., At later times , only ~30% of the replication complexes appear to be active at a given time , as defined by ( − ) strand colocalization with either ( + ) RNA , NS3 , or NS5A ., While both ( + ) and ( − ) RNAs colocalize with the viral proteins NS3 and NS5A , only the plus strand preferentially colocalizes with the viral envelope E2 protein ., These results suggest a defined spatiotemporal regulation of HCV infection with highly varied replication efficiencies at the single cell level ., This approach can be applicable to all plus strand RNA viruses and enables unprecedented sensitivity for studying early events in the viral life cycle . | The stages of the viral life cycle are spatially and temporally regulated to coordinate the infectious process in a way that maximizes successful replication and spread ., In this study , we used RNA in situ hybridization ( ISH ) to simultaneously detect HCV ( + ) and ( − ) RNAs and analyze the kinetics of HCV infection at the single cell level as well as visualize HCV RNAs associated with actively translating ribosomes , markers of viral replication compartment formation , active RNA replication , nucleocapsid assembly , and intracellular virions ., We observed a spatial linkage between sites of viral translation and replication , in addition to replication and assembly ., HCV ( + ) RNAs follow a tight temporal regulation ., They are initially associated with translating ribosomes , followed by a peak of replication that achieves a steady state level ., The remaining HCV ( + ) RNAs are then devoted to virion assembly ., Analysis of HCV ( − ) RNAs revealed that low levels of transient RNA replication occur early after infection prior to the formation of devoted replication compartments and robust replication ., This suggests that HCV synthesizes additional ( + ) and ( − ) strands early in infection , likely to decrease its reliance on maintaining the integrity of the initially infecting ( + ) RNA . | null | null |
journal.pgen.1003388 | 2,013 | Ancient Evolutionary Trade-Offs between Yeast Ploidy States | A central yet poorly understood variable of life is the number of chromosome sets contained within the nucleus of eukaryotic cells ., Ploidy varies throughout the tree of life , with ancient polyploidization events close to the angiosperm 1 , 2 and vertebrate 3 , 4 radiations and among yeasts 5 ., Variation in ploidy states was initially predicted to be neutral as the balance between genes was assumed to be unperturbed 6 ., However , it has recently become clear that ploidy has substantial impacts , defining genome evolution and heredity 7 , controlling organismal development through transient establishment of specialized polyploid cell types 8 and promoting tumor progression 9 ., Despite the biological impact of ploidy differences , the underlying molecular , evolutionary and ecological constraints controlling these remain murky 10 ., Mutational models are based on chromosome set additions increasing the number of mutable sites but masking recessive variation , thereby affecting the emergence , tolerance to and purging of de novo mutations 11 ., Factors such as strength of selection , mutation rate , population size and ratios of deleterious to adaptive and recessive to dominant mutations consequently determine whether a particular ecological context will favor high or low ploidy 10 , 12–14 ., In contrast , cell size models presuppose higher ploidy states to increase cell and organelle volume but to fail to proportionally enlarge surface areas 15 , thereby distorting the balance between transport rates , costs and needs ., In these models , the abundance of beneficial and harmful substances imposes selection for different ploidy states in different environments 15–18 ., Finally , life history models note the intricate interlacing of ploidy variation with alterations of mating , meiosis and sporulation patterns , which originate in the ploidy dependent genetic composition at sex determining loci and the resulting ploidy dependent initiation of dedicated transcriptional programs 19 ., This potentiates co-selection on ploidy and ability to mate , outbreed and sporulate in response to mostly unknown environmental cues ., The unicellular bakers yeast , Saccharomyces cerevisiae , reproduces asexually in stable haploid , diploid and polyploid forms and has emerged as a key model for ploidy research ., Here , we exhaustively mapped the impact of ploidy on the mitotic fitness of S . cerevisiae and its never domesticated relative Saccharomyces paradoxus across wide swaths of their genotypic and phenotypic space ., Influences of ploidy on asexual proliferation in different ecological contexts were found to be the rule rather than the exception with the majority of ploidy effects being well conserved over the 2 billion generations separating the two species 20 ., This demonstrates preservation in the face of considerable genetic drift and large ecological upheavals ., Previous hypotheses of generalizable advantages of haploidy or diploidy in ecological contexts imposing nutrient restriction , toxin exposure and elevated mutational loads were rejected in favor of more fine-grained models of the interplay between ecology and ploidy ., Cell size and mating type locus composition each explained 12 . 5–17% of ploidy effects in the universal reference strain S288c ., To map the impact of ploidy on the capacity for asexual reproduction across the genomic and phenotypic space of the species , 24 S . cerevisiae and 27 S . paradoxus natural isolates ( Table S1 ) were propagated clonally as haploids and MATa/α autodiploids in 33 distinct environments ( Table S2 ) ., Together , these isolates represented >90% of the known genetic 21 and phenotypic 22 variation within these species and encompassed the major populations , geographic origins and source environments ( Table S1 ) ., From >12 . 000 high density population growth curves , we extracted the mitotic fitness components lag ( population adaption time ) , rate ( population doubling time ) and efficiency ( population density change ) of clonal reproduction ( Figure 1A ) ., These measures together encapsulate the capacity of yeast for asexual proliferation , the dominant mode of yeast reproduction in the wild 23 , 24 , and are thus likely to influence yeast fitness substantially in natural contexts ., Considering the complete range of environmental and genetic contexts , the performance of haploids and diploids adhered closely to the 1∶1 null hypothesis expectation of overall equal performance of haploids and diploids ( Figure 1B ) ., The tendency towards similar performance of haploids and diploids was evident for all mitotic fitness components , for both species and for all populations , source habitats and genetic backgrounds ( Figure 1C , Figures S1 and S2 ) ., Hence , considering a wide section of environmental space , we conclude that evolution has failed to establish a decisive asexual reproductive advantage of either haploidy or diploidy ., Despite the absence of a general mitotic advantage of either haploid or diploid genome architecture , ploidy dramatically affected the mitotic capacity in distinct ecological contexts ( Figure 2A ) ., Taking the complete genomic space of the two species into account , significant ( FDR , α\u200a=\u200a0 . 05 ) differences between ploidy states were observed in a vast ( 70% ) majority of all environments , including in optimal conditions ( Figure S3 ) ., Thus , ploidy is more likely than not to affect the asexual proliferation of yeast in any given environmental context ., Considering haploid and haploid strains separately , the radiation into S . cerevisiae and S . paradoxus was a key determinant of phenotype variation , explaining 25 . 2% ( ANOVA F-test , p\u200a=\u200a1 . 2E-82 ) and 14 . 5% ( ANOVA F-test , p\u200a=\u200a1 . 7E-45 ) of the variance in strain pair similarity ., However , the species divergence had essentially no impact on the effect of ploidy on traits , explaining only 2 . 5% ( ANOVA F-test , p\u200a=\u200a7 . 8E-9 ) of the similarity between strains with regards to ploidy-environment interaction ( Figure 2B ) ., In fact , the majority of significant ploidy effects were strikingly evident in both S . cerevisiae and S . paradoxus ( Figure 2A ) ., Thus , despite substantial trait differentiation during the 2 billion generations having passed since species radiation , many of the ploidy effects have remained conserved , although with substantial quantitative variations between species ., To further explore the evolutionary origin of ploidy effects in S . cerevisiae , we estimated the degree to which the historical separation into distinct populations could explain the variation in ploidy effects , population structure being the major determinant of trait variation among S . cerevisiae strains 22 ., However , population structure explained only 9 . 3% of variation in ploidy-environment interactions within S . cerevisiae ., The North American and Malaysian populations showed virtually identical ploidy effects within populations , fully accounting for this explanatory power ( Figure 2B ) ., The later , human enforced separation of S . cerevisiae into clinical , fermentation , lab and wild strains only accounted for a further 1 . 8% of the variation in ploidy effects ( Figure 2B ) ., This is in line with the generally limited explanatory power of human influence on S . cerevisiae trait differentiation 22 ., Taken together , our observations suggest ploidy-environment interactions to have originated in the period of shared evolutionary history of S . cerevisiae and S . paradoxus ., Since the divergence of these species , these ploidy-environment interactions have resisted both natural and human imposed genetic drift and selection , consistent with the action of strong selection ., The quantitative differences between the species are in line with that the strength of selection acting on each type of ploidy-environment interaction , although present in both species , has diverged somewhat during their recent , separate evolution ., Environmental contexts were selected specifically to allow testing of hypotheses on the beneficial effects of diploidy in environments elevating mutational loads and in environments rich in toxic substances and of beneficial effects of haploidy during nutrient restrictions ., Our data failed to support a general mitotic fitness advantage of diploidy in environments associated with elevated mutation rates ., Instead , the type of DNA damage induced appeared to define the relative merits of a haploid and diploid asexual proliferation ., Phleomycin , inducing DNA lesions via a free radical based mechanism , clearly favored diploidy across the genomic range of both species ( Figure 2A , Figure 3 ) ., However , no systematic bias was detected in doxorubicin , intercalating between DNA bases , or in cisplatin , a DNA crosslinker creating adducts between purine residues ., Hydroxyurea , impeding DNA repair by depleting deoxynucleotides , instead strongly favored haploids ., Exposure to some mutagens increases the rate of ploidy switching 25 ., To exclude confounding effects of mating type switching , we therefore quantified the stationary phase DNA content of haploid and diploid populations of five strains in the absence of stress and during Doxorubicin , Hydroxyurea and Cisplatin exposure ., In no case was ploidy switching on the population level observed , although minor ploidy polymorphisms may have emerged in some cultures ( Figure S4 ) ., Opposing ploidy effects were found also during nutrient restriction ( Figure 2A , Figure 3 ) ., Depending on the nitrogen source , nitrogen restriction was either ploidy neutral , or favored either haploids or diploids ., Thus , environments containing tryptophan or leucine as sole nitrogen sources provided advantages for diploidy whereas environments containing phenylalanine or urea as sole nitrogen sources benefitted haploids ., Removal of essential micronutrients , forcing mobilization of internal nutrient storages across organelle surfaces , also alternately favored haploids ( inositol depletion ) or diploids ( zinc , magnesium depletion ) ( Figure 2A ) ., Also during exposure to harmful substances , the merits of ploidy shifted dramatically with the specific toxin encountered and failed to follow any of the hypothesized patterns ., For example , exposure to Li+ strongly favored haploidy whereas no such bias was seen for Na+ ( Figure 2A ) ., This is remarkable given that these alkalic cations are considered to act intracellularly through similar mechanisms and are detoxified through similar cellular processes 26 ., In some cases , for example rapamycin and caffeine exposure , the picture was complicated by ploidy dependent trade-offs between the rate and efficiency of asexual proliferation ., Most notably , when populations were supplied with an excess of nutrients and expanded at their maximal rate , S . cerevisiae haploids tended to reproduce faster asexually but achieved a lower total change in population density and were slower in initiating growth ( Figure 4 ) ., Overall , our data falsified assumptions of generalizable effects of ploidy on mutation tolerance , toxin exposure and nutrient utilization , leading us to argue for more nuanced models based on the molecular architecture of cellular responses to individual ecological factors ., A shift from haploidy to diploidy often enlarges cell and organelle volume through prolonged repression of the G1 cyclin Cln3 which links cell cycle to cell size 27 ., However , given that a roughly spherical form is maintained , such a volume increase is not accompanied by comparable enlargement of surface areas ., Given an excess of nutrients other than glucose and no environmental stress , diploids of the universal reference strain S288c possess twice the cell volume of haploids , but their cell surface area is only 1 . 57 times larger ., Diploids regulate their production of cell envelope proteins to match this distortion 28 , but other protein classes are not as stringently regulated , suggesting a potential mechanism for ploidy-environment interactions ., We reasoned that if ploidy-environment interactions indeed arise as consequences of cell size dependent distortions of volume-to-surface area ratios , then artificial cell size enlargement or reduction should inflict similar environment dependent shifts in asexual reproductive performance ., Testing this prediction , S288c haploid and diploid yeasts artificially designed to have enlarged or reduced cell size through gene deletion 29 ( Table S3 ) , were cultivated in environments favoring either haploidy or diploidy in this particular cognate genetic background ( Figure S5 , Table S4 ) ., Of 24 mitotic fitness traits probed , 17% were clearly ( FDR , α\u200a=\u200a0 . 05 ) size dependent considering both haploids and diploids ( Figure 4A , 4B ) ., Small cells consistently showed shorter lag phase when exposed to rapamycin , a Streptomyces toxin inhibiting the growth-promoting TOR pathway ., Small cells also showed consistently superior proliferation efficiency during exposure to heat , Cu2+ or Li+ , presumably reflecting more efficient utilization of energy ., In S288c growth efficiency during Cu2+ or Li+ exposure , at the relevant pH ( 5 . 8 ) , is almost completely determined by recent gene amplifications of the copper chelating metallothionein CUP1 and Li+ exporter ENA 22 ., Hence , the impact of ploidy and cell size on these traits was deemed likely to depend on copper chelation and lithium efflux respectively ., To test if the ploidy effect on asexual growth efficiency during lithium exposure was indeed coupled to ENA mediated lithium efflux , the three S288c ENA genes ENA1 , 2 and 5 , which derive from a single non-ancestral ENA variant recently introgressed from S . paradoxus into the European S . cerevisiae population and later amplified in tandem in S288c 22 , were deleted in MATα haploids and MATa/α diploids ., Deletion of the ENA genes rendered both haploids and diploids hypersensitive to LiCl ., Remarkably , when measuring the growth efficiency of Li+ exposed ena1Δ2Δ5Δ cells at 30 mM LiCl , causing a roughly similar trait reduction as that of WT cells exposed to 225 mM LiCl , we found the ploidy effect on efficiency to be not only obliterated but actually reversed by removal of the ENA genes ( Figure 4C ) ., Thus , when using the Ena genes for Li+ extrusion , haploids sustain a more efficient growth than diploids , whereas they when forced to rely on the lower capacity Nha1 system for Li+ extrusion 26 are overtaken by diploids ., The superior mitotic efficiency of haploids when exposed to Li+ is conserved throughout S . paradoxus and S . cerevisiae ( Figure 4D ) , regardless of the type and number of ENA genes maintained , suggesting this to be an evolutionary ancient trait ., Interestingly , the slower growth rate of haploids when exposed to Li+ , which appeared to be independent of cell size , was evident also in the absence of ENA genes ( Figure S6 ) ., This disconnection between rate and efficiency of Li+ growth emphasizes the complexity of ploidy effects and the necessity to resolve mitotic fitness into its underlying components when considering the underlying molecular mechanisms ., Diploids ( 2n ) in yeast are naturally heterozygous at the mating-type locus ( α/a ) , whereas haploids ( 1n ) contain only one type of genetic information at this locus , either a or α ., This single genetic difference underlies fundamental differences in life-cycle related phenotypes 30 , and could explain the dramatic effects of ploidy on mitotic fitness components in different environmental contexts ., To separate the effect of mating type from other ploidy effects , we considered S288c diploids that are hemizygous at the mating-type locus , carrying either α or a information ., Together with the normal 2n ( α/a ) , 1n ( a ) and 1n ( α ) strains , these were cultivated in environments favoring either S288c haploidy or diploidy ( Table S4 ) ., Significant ( FDR , α\u200a=\u200a0 . 05 ) differences between the hemizygotic α and a diploids and the normal α/a diploid , were then identified , pointing at cases in which the mating type locus contributed significantly to the trait differences between haploids and diploids ., Most asexual proliferation traits , such as the atypical superior performance of S288c diploids in hydroxyurea , were completely independent of mating type locus composition ( Figure 5A ) ., 12 . 5% of the 24 tested traits were affected by mating type locus composition ., This included the superior growth rate of S288c diploids in conditions of nutrient excess and absence of stress and the superior haploid efficiency of proliferation in the face of a doxorubicin mediated elevation of mutation rates ( Figure 5B–5D ) ., The superior growth rate of diploids following a challenge with the TOR inhibitor rapamycin effect is especially noteworthy given the cell size mediated beneficial impact of haploidy on rapamycin growth lag ( Figure 4A , 4B ) ., Thus , a diploid mating type enabled faster cell cycle progression during rapamycin exposure , whereas a haploid cell size enabled faster cell cycle re-entry in the same conditions ., This underscores the complexity of the interplay between ploidy and environment ., The TOR complexes function as key transcriptional activators of ribosomal gene expression 31 ., Given that the strong and consistent elevation in ribosomal protein mRNA levels in haploids relative to diploids 32 , the role of TOR in ribosomal protein transcription is a likely cause of the here observed ploidy effects ., S . cerevisiae lab strain gametes of complementary mating types mate and diploidize after only a few rounds of haploid clonal reproduction , thereafter maintaining diploid mitosis until nutrients in the local environment are exhausted 33 ., Also in lab strain experimental evolutions , initially haploid populations sometimes end up as diploid through successive chromosome replications without cell division 34 , 35 ., This processes proceeds even when selection is limited through repeated single cell passages 36 ., Thus , the drive towards diploidy has been considered to be deeply ingrained in the genome of yeast lab strains ., Considering a large fraction of yeast genotypic space , we found no overall bias towards superior performance of diploids ., The apparent discrepancy between the general tendency towards diploidization and the distinctly superior mitotic proliferation of haploids in many environments begs explanation ., Yeast life history with frequent and narrow population bottlenecks promotes trait divergence through genetic drift 37 and it cannot be excluded that some of the observed ploidy effects represent non-beneficial traits that became fixed in the common ancestor of S . cerevisiae and S . paradoxus during periods of small population sizes ., Furthermore , the routine approximation of yeast asexual reproduction to fitness 38 may not completely reflect the action of selection ., Natural yeasts spend most of their chronological life time in non-dividing states , meaning a potentially superior fitness influence of viability ., Viability is sometimes enhanced by spore form transitions 39 , necessitating a preceding diploidization 19 ., Conceivably , this could disconnect ploidy effects on asexual reproduction from ploidy effects on overall fitness ., Nevertheless , the frequent conservation of ploidy effects across the 2 billion asexual generations separating S . cerevisiae and S . paradoxus suggests such a decoupling to be unlikely to explain the bulk of the observed effects ., In fact , it implies strong selection to have acted on the ploidy-environments interactions in both these species since the time of their divergence ., This leaves the alternative explanation; that the tendency towards diploidization is not a universal feature of S . cerevisiae in natural habitats ., The recent emergence of yeast population genomics 21 and phenomics 22 has enforced the realization that S . cerevisiae properties vary within surprisingly wide boundaries ., Ploidy preference , varying enormously between yeast species but unstudied over a wider section of the genotypic and ecological space of S . cerevisiae , may be similarly fleeting , as supported by a surprisingly large natural variation in ploidy at micro-ecological scales 40 ., Mutation rates are thought to be independent of ploidy state 41; thus , an increase in DNA content confers a proportional rise in mutational load 42 ., In addition , the almost universal lack of penetrance of gene-disrupting mutations as long as one functional copy remains 37 , 43 causes mutation masking effects , impeding purging of mutations ., Given the overwhelmingly negative nature of mutations , both these effects should favor haploidy in the long run ., However , mutation masking may allow sustained proliferation during short periods of elevated mutation rates , selecting against haploidy in niches where such fluctuations are frequent 44 ., Our data rejected a general asexual advantage of diploidy in environments elevating mutation rates ., Applied doses of mutagens impaired the asexual proliferation of most strains ., However , it cannot be excluded that these costs arose from perturbation of other cellular features , such as transcription , or were associated with drug export or metabolism , or arose from the costs of repairing DNA damage ., Strictly speaking , we cannot tease apart the effects of unrepaired mutations , effects of repairing DNA damage , and effects on other cellular features ., This calls for some caution when interpreting results ., The influence of ploidy may also be strongly dependent on the type and mechanism of DNA damage ., Double stranded breaks disproportionately challenge haploids as repair by homologous recombination , using an extra unperturbed chromosome copy , is by far the most efficient repair mode 45 ., Smaller base lesions , resulting from oxidation and alkylating damage , impose no such requirements 46 ., Haploid and diploid yeast activate different DNA repair pathways in response to replication stress imposed by Mcm4 impairment 47 ., It is conceivable that e . g . the effect of ploidy on hydroxyurea tolerance , also impairing replication , may be due to this differential DNA repair activation ., Increase in ploidy can also result in decreased genome stability due to disproportionate scaling of chromosome segregation components , notably the kinetochore , spindle and spindle pole body 48 ., Such imbalances may fuel the strong effects of ploidy on cellular responses to gross chromosomal rearrangements 32 and affect the tolerance to mutagenic agents ., A natural shift from haploidy to diploidy also alters mating type locus composition , from MATa or MATα to MATa/MATα ., This mediates a shift from haploid to diploid specific transcription programs and from preparedness to mate to readiness to pass through meiosis and later sporulation 49 ., The affected pathways are often pleiotropic , raising the potential for phenotypic hitchhiking of mitotic ploidy effects with effects on mating , meiosis and sporulation ., The shift from axial budding in haploids to bipolar budding in diploids is one potential mechanistic mediator of such pleiotropic consequences 50 , as is the 10-fold increase in transposon production in haploids resulting from induction of the pheromone signalling pathway 28 , 51 ., We found mating type locus composition to account for 12 . 5% of ploidy effects on mitotic properties in S288c ., This included superior haploid asexual reproductive efficiency following exposure to doxorubicin ., Doxorubicin induces double strand breaks which requires repair through homologous recombination or non-homologous end joining ., The latter process is turned off in MATa/MATα S288c diploids through the a1-α2 repression of the NEJ1 transcription factor 52 , 53 , suggesting a likely molecular cause for ploidy dependent doxorubicin tolerance in S288c ., An increase in ploidy often has similar effects on cell size through repression of the G1 cyclin linking cell cycle progression to cell size 27 ., Accordingly , benefits of increased ploidy could arise in toxic environments due to elevated cell volume-to-surface area ratios , reducing uptake of harmful compounds relative detoxification capabilities given that the latter are volume dependent 18 ., Analogously , microhabitats where fitness is constrained primarily by nutrient accessibility may favor haploidy due to the lower volume-to-surface area ratio and enhanced nutrient uptake relative to volume 15–17 ., This presupposes nutrient transport across membranes to be a limiting factor in the utilization of the nutrient , which often , but not always , appears to be the case 15 ., The scope of the current study and the absence of a consistent effect of ploidy on either toxin tolerance or nutrient utilization provide grounds for rejecting both these hypotheses in their most generalized form ., A potential cause of the failure of these hypotheses is the ploidy dependent regulation of cell size in response to environmental cues ., In nutrient rich environments diploid S288c boasts 1 . 57 times the volume of haploids but carbon restriction completely eliminates this difference 15 ., Furthermore , it is doubtful whether substance influx is the sole , or even may , variable affecting asexual reproduction that is altered by cell size ., Both efflux and vacuolar storage often have substantial impacts on yeast proliferation under nutrient restriction and toxin exposure and these may be similarly dependent on volume-to-surface area ratios ., Nevertheless , individual ploidy-environment interactions were sometimes explained by cell size ., In the case of the superior asexual reproductive efficiency of haploids during exposure to Li+ , we traced these effects to the presence of the Ena lithium pumps ., Given the enormous influence of the ENA locus on asexual reproductive efficiency in lithium environments 22 and the belief that ATP driven pumping of Li+ by Ena proteins completely controls Li+ efflux at intermediate H+ 26 , this was not entirely surprising ., In absence of the Ena transporters , yeast is forced to rely on Nha1 for alkali metal efflux , a pump that has a vastly lower capacity at pH 5 . 8 as it is driven by proton influx 26 ., Interestingly , the more efficient growth of haploids during Li+ exposure was not only obliterated by removal of the ENA genes , but reversed , now favoring diploids ., This suggests that ploidy has reverse impacts on the Nha1 and the Ena systems , illustrating how a simple molecular shift can completely alter the relative merits of haploidy and diploidy in a particular environmental context ., This is consistent with a recent finding that adaptive mutations emerging and driving towards fixation in evolving laboratory populations have different effect sizes when reconstituted individually in haploid and diploid genomic contexts 54 ., Ena2 , the Ena variant with highest affinity for Li+ , appears to be largely unregulated and expressed at basal levels 55 , suggesting that the density of Ena2 in the diploid membrane , which presumably has a higher surface area , may be lower than the density in the haploid membrane ., This may explain the ploidy effects ., Although alterations in cell volume-to-surface area ratios may mediate many cell size dependent ploidy-environments interactions , it should be noted that also organelle volume-to-surface area ratios fluctuate as a function of cell size and environmental context ., Expansions and fragmentations of yeast vacuoles 56 , and expansion of the nucleus 57 are well documented examples ., Furthermore , a host of other biochemical and regulatory properties also depend on cell size 15 , 29 , such as silencing at some subtelomeric regions via unknown posttranscriptional mechanisms 58 ., All these may contribute to ploidy-environment interactions affecting mitotic properties ., Yeast mitotic properties in different environmental contexts also tend to be highly polygenic 59 , 60 , increasing the likelihood that detected ploidy effects may be composites of cell size , mating type and DNA content influences ., This enhances the challenge of molecularly decoding ploidy dependent traits and may explain why 70% of S288c ploidy effects could not be accounted for by considering cell size and mating type individually ., Overall , our findings revealed an unsuspected prevalence of ploidy effects in yeast and suggested a dynamic interplay between ploidy and environment , involving evolutionary trade-offs of surprisingly ancient origin and diverse molecular roots ., 24 S . cerevisiae and 27 Saccharomyces paradoxus isolates , corresponding to known yeast populations , geographic origins and source environments ( Table S1 ) , were isolated as described 21 ., Following deletion of URA3 ( KanMX ) and HO ( HygMX ) , mating and sporulation , haploid ( MATa and MATα ) and autodiploid ( MATa/MATα ) were obtained 61 and long-time stored at −80°C in 20% ( w/v ) glycerol ., Strains were subjected to high throughput phenotyping by micro-cultivation in 33 environments essentially as previously described 62 , 63 ., A complete list of environments can be found in Table S2 ., Strains were inoculated in 350 µL of Synthetic Defined ( SD ) medium ( 0 . 14% yeast nitrogen base , 0 . 5% ammonium sulfate and 1% succinic acid; 2% ( w/v ) glucose; 0 . 077% Complete Supplement Mixture ( CSM , ForMedium ) , pH set to 5 . 8 with NaOH or KOH ) and incubated for 48 h at 30°C ., For experiments where the removal of a specific nutrient was studied , the pre-culture was performed in absence of this nutrient in order to deplete intracellular storages ., For experiments where alternative nitrogen sources were used , two consecutive pre-cultures were performed , the first in limiting concentrations of ammonium , 29 µg N/mL , in order to avoid excessive nitrogen storage , the second replacing ammonium with the indicated nitrogen source in amounts corresponding to an equivalent number of nitrogen atoms ., Except for the nitrogen source indicated and 20 mg/L uracil , which cannot be used as sole nitrogen source 22 , no other nitrogen was supplied in these experiments ., For experimental runs , precultures were diluted 35× to an OD of 0 . 03–0 . 15 in 350 µL of SD medium and cultivated for 72 h in a Bioscreen analyzer C ( Growth curves Oy , Finland ) ., Optical density was measured using a wide band ( 450–580 nm ) filter ., Incubation was at 30 . 0°C ( ±0 . 1°C ) with ten minutes preheating time ., Plates were subjected to shaking at highest shaking intensity with 60 s of shaking every other minute ., OD measurements were taken every 20 minutes ., The rate ( population doubling time ) , lag ( population adaptation time ) and efficiency ( total change in population density ) of asexual reproduction were extracted from high density growth curves and log2 transformed 62 , 63 ., Relative mitotic fitness components for each strain and environment , LSCij , were calculated by normalization of each measurement to an internal ( WT ) standard ( haploid S288c , MATα , n\u200a=\u200a8 ) as:wtkj is the trait measure of the kth measurement of the wild type for trait j , xij is the measure of strain i for trait j and r indicates the run ., To maintain directionality between the mitotic fitness components , the measure for proliferation efficiency was inverted ., Note that the lag measures generally should be treated with caution due to its higher sensitivity to bias ., For example , it cannot be excluded that some early growth is misclassified as a lag , due to the cell density increase being below the threshold of detection ., To compare haploid and diploid asexual proliferative capacity , a mean of the two mating types ( each n\u200a=\u200a2 ) was used to derive a single measure of haploid performance ., This was compared to that of the diploid ( n\u200a=\u200a4 ) ., For S288c , a substantially higher number of MATα haploids ( n\u | Introduction, Results, Discussion, Materials and Methods | The number of chromosome sets contained within the nucleus of eukaryotic organisms is a fundamental yet evolutionarily poorly characterized genetic variable of life ., Here , we mapped the impact of ploidy on the mitotic fitness of bakers yeast and its never domesticated relative Saccharomyces paradoxus across wide swaths of their natural genotypic and phenotypic space ., Surprisingly , environment-specific influences of ploidy on reproduction were found to be the rule rather than the exception ., These ploidy–environment interactions were well conserved across the 2 billion generations separating the two species , suggesting that they are the products of strong selection ., Previous hypotheses of generalizable advantages of haploidy or diploidy in ecological contexts imposing nutrient restriction , toxin exposure , and elevated mutational loads were rejected in favor of more fine-grained models of the interplay between ecology and ploidy ., On a molecular level , cell size and mating type locus composition had equal , but limited , explanatory power , each explaining 12 . 5%–17% of ploidy–environment interactions ., The mechanism of the cell size–based superior reproductive efficiency of haploids during Li+ exposure was traced to the Li+ exporter ENA ., Removal of the Ena transporters , forcing dependence on the Nha1 extrusion system , completely altered the effects of ploidy on Li+ tolerance and evoked a strong diploid superiority , demonstrating how genetic variation at a single locus can completely reverse the relative merits of haploidy and diploidy ., Taken together , our findings unmasked a dynamic interplay between ploidy and ecology that was of unpredicted evolutionary importance and had multiple molecular roots . | Organisms vary in the number of chromosome sets contained within the nucleus of each cell , but neither the reasons nor the consequences of this variation are well understood ., We designed yeasts that differed in the number of chromosome sets but were otherwise identical and mapped the consequences of such ploidy variations during exposure to a large palette of environments ., Contrary to commonly held assumptions , we found ploidy effects on the mitotic reproductive capacity of yeast to be the rule rather than the exception and to be highly evolutionarily conserved ., Furthermore , our data rejected previously contemplated hypotheses of generalizable advantages of haploidy or diploidy when cells face nutrient starvation or are exposed to toxins or increased mutation rates ., We also mapped the molecular processes mediating ploidy–environment interactions , showing that cell size and mating type locus composition had equal explanatory power ., Finally we show that ploidy effects can be mechanistically very subtle , as a designed shift from one plasma membrane Li+ transporter to another completely altered the relative merits of having one or two chromosome sets when exposed to high Li+ concentrations ., This complex and dynamic interplay between the number of chromosomes sets and the fluctuating environment must be taken into account when considering organismal form and behavior . | organismal evolution, population genetics, ploidy, quantitative traits, microbiology, model organisms, eukaryotic evolution, biology, trait locus, phenotypes, heredity, genetic screens, genetics, yeast and fungal models, saccharomyces cerevisiae, evolutionary biology, genetics and genomics | null |
journal.pbio.1000546 | 2,010 | Evolutionary Relationships of Wild Hominids Recapitulated by Gut Microbial Communities | The mammalian digestive tract is sterile at birth but is soon colonized by bacteria that typically derive from the mother 1–3 ., In the absence of any subsequent alterations or additional colonizations , strict parental inheritance would result in a pattern in which the constituents and composition of the microbial flora would co-diversify with and ultimately mirror the evolutionary relationships of their hosts ., Such a situation has been observed in some bacteria within the digestive tract , such as Helicobacter pylori , which is present in the stomachs of about half of the human population and whose patterns of divergence closely follow those of their human hosts 4 ., Numerous internal and external factors , including diet , geography , host physiology , disease state , and features of the gut itself , contribute to the community composition of the gut microbiota 5–9 and can result in discordance with the host phylogeny ., Despite the wide variation among individuals , the gut microbiotae of members of the same species are often more similar to one another than to those of other species ., But above this level of organization , the composition of these microbial communities is thought to assort according to the broad dietary habits of their hosts 6 , 10 ., Based on very limited samplings of nonhuman primates , mostly from captive individuals , conspecifics sometimes retain very similar microbial communities ( e . g . , Hymadryas baboons ) , but sometimes do not ( e . g . , western lowland gorillas ) ., And in a previous phylogenetic analysis of mammals based on their gut microbiotae , the great apes were interspersed in multiple clades along with distantly related species 6 , 10 ., For example , humans , bonobos , and two of the three gorilla species landed in a large “omnivore” clade along with lemurs , an elephant , and an armadillo , whereas the chimpanzees and orangutans grouped with a flying fox in a divergent clade 10 ., From such isolated cases of displaced or zoo-raised hosts , it is difficult to extract the degree to which host and environmental factors shape the primate gut microbiota: both factors are certainly important , but their relative contributions cannot be established based on previous sampling ., To address questions pertaining to the stability and variation in the great ape gut microbiota over evolutionary timescales , we performed high-coverage sequencing of the small subunit ribosomal RNA genes 11–13 present in the feces of apes collected in their native ranges ., The samples from these wild-living hominids included eastern and western lowland gorillas , bonobos , and three subspecies of chimpanzees 14–17 , as well as two human hosts from different continents ., This sampling provided a more comprehensive and less biased view of bacterial species diversity and abundance within the primate distal gut , and revealed that the relationships among microbial communities parallel the host-species phylogeny ., Our results indicate that evolutionary changes in host physiology that occurred during the divergence of great apes have been the dominant factor in shaping the distal gut microbial community present in each host species ., The gut microbiotae of these hosts encompass one archaeal and 18 bacterial phyla , of which five ( Actinobacteria , Bacteroidetes , Firmicutes , Proteobacteria , and Verrucomicrobia ) were present in all samples ., Several phyla not typically observed in gut microbiota of primates , including Euryarchaeota , Acidobacteria , Fibrobacteres , Lentisphaerae , Planctomycetes , and candidate phylum TM7 , were recovered at very low relative frequencies ( <10−3 ) from at least nine hosts ., In addition , five bacterial phyla ( Chlamydiae , Chloroflexi , Deferribacteres , OP10 , and Gemmatimonadetes ) were detected in only one or few hosts ., Contributing most to these rare variants was chimpanzee BB089 , which had the highest phylum-level diversity of any sample and harbored four of these five uncommon phyla ( Table S1 ) ., The three most dominant phyla were Firmicutes , Proteobacteria , and Bacteroidetes , which together constituted over 80% of the reads identified in every sample ., Although divergent mammals can harbor broadly similar gut microbiotae at the level of bacterial phylum 7 , 10 , the two species of gorilla differed from those of other great apes in the relative frequencies of the dominant phyla ., Firmicutes was numerically dominant in all great apes but was less common than Proteobacteria and Bacteroidetes in both species of gorilla ., Among other phyla represented in all hosts , Actinobacteria was common but only occurred at frequencies of greater than 10% in two of the five bonobos and in a single chimpanzee ( CP470 ) , and Verrucomicrobia , usually at frequencies of only 1%–10% , constituted approximately 20% of the microbiota of one human ( KS477 ) ., To determine whether great ape species can be distinguished based on the diversity of microbes in their fecal samples , we first performed a phylogenetic analysis of the phylum-level diversity within their gut microbiota ., For this analysis , we contructed phylogenetic trees based on the abundance in each sample of pyrotags that were classified to phylum ( see Materials and Methods and below for results based on species-level microbial diversity ) ., Despite variation in the distribution and abundance of numerous microbial phyla , these phylum-level phylogenies did not resolve any of the ape species as discrete groups ( Figure S1 ) ., There were 160 most parsimonious trees , and no clade recovered had greater than 70% bootstrap support ., This result is due to both the sporadic occurrence of certain phyla among individual members of the same great ape species and the phylum-level diversity present in the microbiota of chimpanzees , which broadly overlaps that of the other great apes ., The majority of the variation in the microbiotae of the great ape hosts is represented as unique pyrotags recovered from a single sample , indicating that differentiation in the gut microbiota among hosts may occur at lower taxonomic levels ., Although these 16S rDNA sequences are indicative of a broad range of species-level bacterial diversity in these samples , the source , relevance , and reproducibility of this “rare biosphere” has recently been questioned 18–20 ., To assess how experimental factors might contribute to the contents of the rare biosphere , we performed a high-coverage technical replicate ( WE464R ) on an independent preparation of the fecal sample from chimpanzee WE464 ., Even with sequencing to 3 . 5 times the depth of the initial sample ( 51 , 648 versus 14 , 762 reads ) , there were no identical matches for approximately 30% of the reads in the original sample , but when allowing for up to 0 . 5% sequence divergence between reads ( i . e . , no more than a single one-nucleotide mismatch or indel ) , this proportion shrank to less than 10% ., Therefore , to assemble the most biologically robust segment of our entire 1 , 107 , 714-pyrotag dataset , we grouped sequencing reads by applying a 99 . 5% identity threshold to correct for most potential sequencing artifacts ., We have noted previously that thresholds higher than 97% will inflate richness estimates using amplicon pyrosequencing 20; however , the approach we take here will be minimally impacted by this artifact ., More importantly , application of this high threshold improves the likelihood that clustered pyrotags belong to the same bacterial species , whereas the conventional criterion of 97% sequence identity often unites bacteria typed to different taxonomic groups 21 , 22 ., To determine the degree to which the gut microbial communities present in these great apes are similar in the frequencies of their constituent microbial species , we retained only those 99 . 5% operational taxonomic units ( OTUs ) detected in two or more host samples ( since unique OTUs are not phylogenetically informative and provide no information about evolutionary relatedness ) ., The set of reproducible 99 . 5% OTUs contained a total of 1 , 017 , 478 reads that formed a total of 8 , 914 microbial phylotypes ( hereafter referred to as “species” ) , which were used to examine the fine-scale taxonomic structure and similarities of the great ape gut microbiotae ( Figure 2 ) ., Whereas our analyses of phylum-level microbial diversity were not sufficiently grained to differentiate great ape species based on their microbiota ( a similar limitation encountered by Ley et al . 10 when only 100–200 bacterial sequences were sampled from each host ) , the assemblage of microbial species ( that manifest as reproducible 99 . 5% OTUs ) discriminated the individual primate hosts and assorted them into taxonomic groupings ., For example , the two humans shared relatively few phylotypes with other great apes , and both species of gorillas shared high frequencies of proteobacterial phylotypes and very low frequencies of Firmicutes species that were present in the majority of chimpanzee samples ., This result indicates that deep sampling of the microbiota is necessary to fully recover the evolutionary signal in gut microbial community data ., The number of reproducible microbial species recovered from individual hosts ranged from 265 ( in African human KS477 ) to 3 , 247 ( in chimpanzee WE458 , the host for which we obtained the highest number of reads ) ., The highest frequency attained by an individual bacterial species was 27 . 7% for a phylotype classified as Megasphaera ( Firmicutes: Clostridium ) in the sample from the African human ., To conduct a phylogenetic analysis of the occurrence and frequencies of species present in the fecal microbial communities in the primate hosts , we treated each microbial species as an individual standard data character assigned to one of six possible states that correspond to order-of-magnitude differences in the normalized frequency of each microbial species in each ape host sample ., This approach is similar to that typically used with morphometric data , where , for example , femur length might be treated as a standard character coded with a handful of possible states ranging from very small to very large ( e . g . , the multi-log-difference size range from mouse to elephant ) ., The aggregate character matrix of several morphometric characters ( or , in our case , the frequencies of the various members of a microbial community ) can then be analyzed using traditional phylogenetic techniques ., The 8 , 914-species , six-state data matrix was subjected to a heuristic maximum parsimony tree search , as performed previously for the phylum-level tree phylogeny , with 1 , 000 pseudo-replicates used to assess bootstrap support ., Because microbial species are defined as reproducible OTUs clustered at 99 . 5% sequence identity , all 8 , 914 characters are parsimony-informative , and the phylogenetic analysis recovered a single maximum parsimony tree ( p-score\u200a=\u200a49 , 475 ) ., The unrooted maximum parsimony tree exhibited a species-level topology that was completely congruent with the unrooted mtDNA topology of the hosts ( Figure 3 ) ., Moreover , these groupings were supported by high bootstrap values ( 98%–100% for each ape species , with the exception of the chimpanzee clade , which reached only 68% bootstrap support ) ., There are more than 2 , 000 , 000 possible unrooted topologies for a ten-taxon phylogenetic tree ., Therefore , if one constructs a tree with two humans , two chimpanzees , two bonobos , two eastern lowland gorillas , and two western lowland gorillas , the chance of randomly generating a tree that is entirely congruent with the species tree in placing each species with its conspecific , and also placing the two gorilla species as sister groups and the chimpanzees and bonobos as sister groups ( as in Figure 3 ) is less than 1/2 , 000 , 000 ., Based on the compositions of the distal gut microbial communities from hosts living in their natural environments , we were able to discriminate species of great apes ., The topological concordance between the species-level branching orders obtained for hosts and their microbiotae shows that over evolutionary timescales , host phylogeny is the overriding factor determining the microbial composition of the great ape gut microbiota ., This recapitulation of the species relationships in the frequencies of the microbial constituents of their distal gut communities contrasts with previous notions that diet is the most important factor governing the grouping of gut microbiotae within primates 6 , 10 ., This new view of great ape microbiota evolution emerged as a consequence of the sampling depth , which allowed the recovery of large sets of evolutionarily informative phylotypes ., This allowed the application of standard parsimony-based phylogenetic approaches that were based on the frequency of each microbial species shared among hosts ., Previous studies of gut microbiotae that surveyed only on the order of 100 sequences per sample 10 , 23 , 24 could not accurately gauge either the diversity present in complex microbial communities or the relative abundance of the constituent species ., Given the species complexity within the distal gut microbiota , it is necessary to obtain more than 104 reads per host to accurately access the relationships among divergent microbial communities ., However , recent advances in sequencing methodologies render this number of reads both technically and economically feasible ., The fact that the gut microbial community phylogeny matches the great ape species phylogeny is not readily attributable to factors other than the evolutionary diversification of hosts ., For example , the broad geographic range of chimpanzees , as well as the intercontinental distance separating our sampled humans , establishes that geographic proximity is not a major factor in the clustering of microbial communities by host species ., Likewise , chimpanzees and gorillas within the same locale exhibited phylogenetically distinct gut microbial communities ., That the composition of gut microbiotae assorts to species despite their geographic locations suggests that similarities in local factors , such as those that relate to diet , do not explain the close correspondence between host phylogeny and microbial community composition ., To further evaluate whether host species differentiate according to diet , we examined the populations of chloroplast sequences within each fecal sample ., Although the diversity of chloroplasts serves as an indicator only of plant diet at the time of sampling , there was no clear indication that the great ape species ( except for G . beringei ) have widely different diets or that the diets of great apes structure according to host phylogeny ( Figure S2 ) ., As evident from the differences in relative branch lengths between the mtDNA ( Figure 3A ) and microbial community ( Figure 3B ) trees , it is clear that the degree of genetic differentiation between hosts does not fully account for the variation in great ape gut microbiota ., The host phylogeny signal that we uncovered can be masked by factors occurring on more proximate timescales ( such as diet , geography , or health status ) ., Only by conducting a phylogenetic analysis of communities that have been more deeply sampled is it possible to detect this signal ., To assess the degree to which differences in gut microbiota reflect the genetic distance between hosts , we compared the amount of variation assigned to the terminal branches of the tree ( i . e . , those leading to individual hosts ) relative to that encompassed in the seven internal branches that differentiate the five great ape species ( grey branches in Figure 3 ) ., The species-discriminating branches together represent 73% of the total genetic distance present in the mtDNA phylogeny , but only 7% of the total distance in the tree based on microbial communities ., This contrasts with the situation for individual hosts , whose branch lengths together constitute 70% of the distance in the microbial tree but encompass only 11% of the total genetic distance ., This disparity reflects the broad variation in microbial communities among members of the same species , as has already been observed in humans 5–7 , 25–28 ., Next , to discount the effects of individual variation , we calculated the correlation coefficient between the relative branch lengths of the seven internal branches in the microbial community tree and the corresponding distances in the mtDNA tree ., Despite the congruence in branching orders , the branch lengths in the mtDNA tree explain only about 25% of the variation in the microbial community tree ., This indicates that gut microbiotae , although diverging in a manner consistent with vertical inheritance , are not changing in a strict time-dependent fashion that reflects the degree of genetic divergence among hosts ., The difference in branch length indicates that individual-level variation in microbial community structure is extensive relative to between-species variation ., Our analysis indicates that host phylogeny has a major role in the diversification of distal gut microbial communities in great apes , a conclusion that can become apparent only when sampling is adequate for robust phylogenetic and evolutionary analyses of microbial species compositions ., Numerous studies have applied UniFrac and related approaches to establish the relationships among microbial communities derived from a wide range of hosts and environmental sources 29–33 ., Despite the highly supported tree that we obtained by parsimony analysis , subjecting our dataset to UniFrac did not recover a tree that matches the host-species phylogeny ( Figure S3 ) ., Unlike parsimony , UniFrac relies on an input tree to specify the evolutionary relationship among bacterial taxa to infer the similarity among microbial communities ., However , for a large dataset with nearly 9 , 000 characters , ensuring the correct inference of tree topology and branch lengths is difficult ., The task of inferring an input tree is all the more problematic because of the relatively short and highly variable sequencing reads that are generated for most metagenomic studies ., The quality of multiple sequence alignment , which is critical for inferring the guide tree , is greatly impacted by the limited read length , the level of sequence variation , and the propensity towards indel sequencing errors ., This problem was almost entirely eliminated from our parsimony analysis ( of species abundance data ) by performing multiple sequence alignments on sets of reads assigned to a particular taxonomic class , not the entire dataset ., Furthermore , when calculating pair-wise sequence identities among reads typed to the same class , indel sequencing errors present in taxonomically different reads are ignored ., Since the V6 region has previously been shown to have low phylogenetic congruency with full-length small subunit ribosomal RNA topologies 34 , the described methods based on species abundances and community compositions serve as an alternative and complementary approach for analyzing pyrotag data ., With the availability of methods that allow the scrutiny of microbial diversity and community structure at finer levels , the challenge now is to determine how best to characterize each specific environment in order to extract the relevant biological information about its constituents ., In the present study , we found that sampling at levels of greater than 10 , 000 reads per sample , the application of stringent cutoffs for species identity , and the focus on parsimony-informative characters helped resolve host phylogeny as the major determinant of distal gut microbial communities in great apes ., Ape fecal samples used in this study were selected from an existing bank of previously collected specimens 14–17 ., All samples except one ( GM173 ) were collected from wild-living , non-habituated apes at remote forest sites in Cameroon , the Central African Republic , and the Democratic Republic of the Congo ( DRC ) ., Sample GM173 was obtained from a habituated male chimpanzee ( Ch-045 ) in Gombe National Park , Tanzania ( Figure 1 ) ., In the field , fecal samples were identified to be of likely chimpanzee , gorilla , or bonobo origin by experienced trackers; however , species and subspecies origins were subsequently confirmed in the laboratory by mtDNA analysis ., This genetic analysis revealed a limited number of initially misidentified specimens from other mammal species , including a handful of samples that were of human origin ., One such sample ( KS477 ) from an unknown individual in the DRC was included in this study ., In addition , a fecal sample was supplied by a human male residing in Tucson , Arizona ( United States ) ., All fecal samples were collected , stored , and shipped in RNAlater ( Ambion ) ., Time , date , and collection site were recorded for each sample ., Samples were shipped at ambient temperatures but subsequently stored at −80°C ., DNA was extracted from 200-µl aliquots of thawed fecal samples by spin-column filtration using the QIAamp DNA Stool Kit ( Qiagen ) , following the manufacturers protocol for isolating DNA for pathogen detection ., DNA was quantified on a Qubit fluorometer ( Invitrogen ) and subjected to PCR amplification of the 16S rDNA region spanned by primers 926F ( 5′-aaactYaaaKgaattgacgg-3′ ) and 1492R ( 5′-tacggYtaccttgttacgactt-3′ ) ., Amplicons encompassed the V6 region , which was selected because of prior use and high level of variability 13 , 20 , 34–36 ., To multiplex amplicons for inclusion on a single sequencing run ( 454 Life Sciences/Roche ) , the appropriate 454 Life Sciences adaptor sequence and a unique three- or four-nucleotide sequence tag ( barcode ) were added to the 5′ end of the forward and reverse 16S amplification primers ., For each primer pair , PCR was performed in triplicate and pooled to minimize PCR biases that might occur in individual reactions ., Each 50-µl reaction consisted of 1 . 25 units of Taq ( GE Healthcare ) , 5 µl of supplied 10× buffer , 0 . 25 µl of 10 mM dNTP mix ( MBI Fermentas ) , 1 . 5 µl of 10 mg/ml BSA ( New England Biolabs ) , 0 . 5 µl of each 10 µM primer , and 40 ng of template DNA , and proceeded at 95°C for 3 min; followed by 25 cycles of 95°C for 30 s , 55°C for 45 s , and 72°C for 90 s; followed by a final extension at 72°C for 10 min ., Amplification products were purified on MinElute PCR columns ( Qiagen ) and quantified ., To obtain a similar number of reads from each sample , amplicons were mixed in equal concentrations prior to pyroseqencing ., Emulsion PCR and sequencing were performed using a GS-FLX emPCR amplicon kit ( 454 Life Sciences/Roche ) , following the manufacturers protocols ., Pyrosequencing proceeded from the barcode at the 5′ end of the 926F primer ., To confirm species and differentiate individual hosts , DNA samples were also tested with primers designed to amplify three hypervariable regions of the D-loop of the great ape mitochondrial genome: HV1 ( nucleotides 15997 to 16498 ) , and HV2 and HV3 ( nucleotides 16517 to 607 ) ., Amplified PCR products were treated with exonuclease I and calf intestinal phosphatase , and directly sequenced from both ends using the amplification primers on an ABI 3700 sequencer ( Applied Biosystems ) ., Sequences were assembled in Sequencher ( Gene Codes Corporation ) and compared to the published mtDNA sequences of great apes ., In addition , we used unambiguous polymorphisms to confirm that each sample came from a different individual ., Pyrosequencing flowgrams were converted to sequence reads using software provided by 454 Life Sciences/Roche ., Reads were end-trimmed with LUCY 37 , using an accuracy threshold of 0 . 5% per base error probability ., Reads lacking exact matches to a recognizable barcode and primer sequence were removed from the dataset , leaving a total of 1 , 292 , 542 reads ( elsewhere referred to as “pyrotags” ) out of the original total of 1 , 501 , 806 reads ., Reads were assigned to individual samples based on identifying barcode sequences ., Barcode and primer sequences were removed from the 5′ end of each read , and the taxonomic origin of each read was established using RDP Classifier ., For quality filtering , we excluded the reads that, ( i ) were shorter than 150 nucleotides in length ,, ( ii ) received bootstrap support for class assignment lower than 70% ( based on RDP Classifier ) ,, ( iii ) mapped to the incorrect region of 16S gene , or, ( iv ) were of chloroplast origin ( based on RDP Classifier ) ., This final filter removed more than 99% of the Cyanobacteria reads , leaving only one read that was assigned to genus GpVII ., For sequences classified as Archaea , we required reads to start at position 844–850 ( relative to the reference sequence from RDP Aligner ) ; for sequences classified as Bacteria , we required the reads to start at position 851–857 ., For the 1 , 107 , 714 pyrotags that could be assigned to a taxonomic class , we performed all pair-wise comparisons to identify unique sequence types ., If two otherwise identical reads differed in length , they were trimmed to the same length ., We used RDP Aligner to perform multiple sequence alignments of all unique sequence types ., Based on these alignments , we calculated the percent identity of all pairs typed to the same class ., Terminal gaps in the 5′ or 3′ end of the alignment were excluded when calculating percent identities ., The clustering step was done using the MCL ( Markov Clustering ) algorithm with the inflation value set to 1 . 5 38 ., 99 . 5% OTUs were partitioned into two sets: unique ( present in one sample ) and shared ( present in at least two samples ) ., Frequencies of shared OTUs were visualized via heatmaps generated in R and used for subsequent analyses ., To establish the degree to which the gut microbiotae of samples were similar with respect to the compositions of their constituent microbes , we constructed phylum-level and species-level phylogenies of hosts based on the frequencies of taxonomically assigned OTUs in their gut microbial communities ., Character matrices based on all reads were converted to phylogenetic trees using a parsimony-based approach ., Each character corresponds to a taxonomically assigned OTU whose frequency in each sample has been normalized by coding with one of six ordered states reflecting log-unit differences in its occurrence , with a OTU absent from a sample coded as state 0 ., Given the range in the occurrence of each OTU across samples ( from 0 to 83 , 840 at the phylum level ) , this resulted in six-state data matrix , which was then subjected to a heuristic maximum parsimony tree search using PAUP version 4 . 0b10 , using default settings ., Characters were considered to be ordered , such that transitions between distant states ( i . e . , samples having very divergent frequencies of a particular phylotype ) were more costly than between similar states ., To produce the tree for input to UniFrac , we took the longest read within each OTU as the representative for multiple sequence alignment and generated alignments using RDP Aligner ( applying the Bacteria model since only ten of the 8 , 914 OTUs represented Archaea ) ., The resulting alignment contained 554 aligned nucleotide sites , and the tree relating the 8 , 914 OTUs was inferred in FastTree version 2 . 1 . 1 39 , 40 ., Tree topology and sample information were uploaded to the Fast UniFrac Web server 41 for the clustering analysis , using the weighted and normalized options to account for differences in OTU abundance and read depth among samples . | Introduction, Results, Discussion, Materials and Methods | Multiple factors over the lifetime of an individual , including diet , geography , and physiologic state , will influence the microbial communities within the primate gut ., To determine the source of variation in the composition of the microbiota within and among species , we investigated the distal gut microbial communities harbored by great apes , as present in fecal samples recovered within their native ranges ., We found that the branching order of host-species phylogenies based on the composition of these microbial communities is completely congruent with the known relationships of the hosts ., Although the gut is initially and continuously seeded by bacteria that are acquired from external sources , we establish that over evolutionary timescales , the composition of the gut microbiota among great ape species is phylogenetically conserved and has diverged in a manner consistent with vertical inheritance . | The microbial communities that inhabit the gastrointestinal tract of humans and other mammals are complex , dynamic , and critical to both health and disease ., The composition and constituents of these communities are influenced by multiple factors such as host diet , geography , physiology , and disease state ., Given the central role of the gut microbiota in the physiology of the host , it is important to determine whether it is predictable and substantially determined by the host , or variable and largely determined by the external environment ( including diet ) experienced by the host ., A valuable way of determining the relative contributions of such factors is by comparing gut microbial communities in closely related host species ., Applying a high-throughput sequencing approach , we profiled the distal gut microbiotae of great ape species sampled in their native ranges and then employed a parsimony-based analysis of phylogenetically informative phylotypes ( i . e . , bacterial taxa residing in multiple individuals ) to determine the relationships among the diverse microbial communities ., Our analyses revealed a clear species-specific signature of microbial community structure ., Moreover , the pattern of relationships among the five great ape species ( Homo sapiens , Pan troglodytes , P . paniscus , Gorilla gorilla , and G . beringei ) inferred from their fecal microbial communities was identical to that inferred from host mitochondrial DNA , indicating that host phylogeny shapes the gut microbiota over evolutionary timescales ., It seems after all that you are not what you eat . | evolutionary biology/human evolution, microbiology/environmental microbiology, evolutionary biology/microbial evolution and genomics | Although bacteria are continually acquired over the lifetime of an individual, the phylogenetic relationships of great ape species is mirrored in the compositions of their gut microbial communities. |
journal.pgen.1002642 | 2,012 | Mu Insertions Are Repaired by the Double-Strand Break Repair Pathway of Escherichia coli | Transposable elements drive genome evolution in many ways – increasing DNA content , rearranging and mutating genes , as well as altering gene regulation 1 ., Temperate phage Mu has played a pivotal role in our current understanding of how movable elements move 2 ., A unique aspect of Mu is that , depending on the phase of its life cycle , it moves using either replicative or non-replicative modes of DNA transposition 3 ., Most of our knowledge of Mu transposition is derived for the replicative pathway , where during lytic growth , Mu amplifies its genome by repeated transposition-replication events which exploit the host replication apparatus 4 , 5 ., In vitro experiments have established that in this pathway , the Mu transposase ( MuA protein ) mediates single-strand cleavages at Mu ends followed by strand transfer of the cleaved ends into target DNA; the latter reaction is greatly assisted by MuB protein ( Figure 1 ) ., The resulting branched strand transfer joint is resolved by target-primed replication , which is initiated by the PriA primosome and completed by the Pol III holoenzyme , and results in duplication of the Mu genome after every round of integration ., At the end of the lytic cycle , Mu genomes are packaged into phage heads such that they include host sequences ( flaps ) from both sides of a Mu insertion ., The non-replicative pathway of Mu transposition is only used when progeny phage infect new hosts 6 , 7 , 8 ., Along with Mu DNA , the phage also inject into the host the phage N protein , which binds at the termini and converts the linear Mu genome into a non-covalently closed supercoiled circle 9 , 10 , 11 ., Integration of the infecting Mu into the host genome follows the same initial nick-join steps of transposition established for the replicative mechanism in vitro; however , instead of target-primed Mu replication , the host flaps are resected and the gaps are repaired by unknown mechanisms 12 ( Figure 1 ) ., Flap resection has not yet been demonstrated in vitro ., This reaction is dependent in vivo on the cryptic endonuclease activity harbored within the C-terminal domain of the transposase MuA ( designated MuANuc in this study ) , as well as on the chaperone protein ClpX 13 , 14 ., ClpX is known to play an essential role during Mu replication , remodeling the Mu transpososome and enabling its transition to a replisome 5 , 15 ( Figure 1 ) ., The alternative choices for resolving the transposition intermediate , i . e . repair versus replication , must involve additional phage and host factors whose identity is not yet established ., The current study was undertaken to identify host factors involved in the repair of Mu insertions during the non-replicative infection pathway ., To do so we used the Keio Collection , which is a set of 3 , 985 precisely defined , single-gene deletions of all nonessential genes in Escherichia coli K-12 16 , and screened for mutants defective in recovery of Mu::Cm insertions ., Among the several mutants that gave a poor yield of CmR integrants , a majority of those that allowed Mu entry showed normal integration and replication of wild type Mu ., By using two additional phage variants to re-screen/re-test in order to eliminate those defective in maintenance of a stable prophage state , we narrowed the search to a small subset of the mutants ., Included among these were mutants in the homologous recombination pathway - recA , recB , recC ., Two mutants - priA and dnaT – were defective in Mu replication as expected , but were unexpectedly defective in the recovery of insertions despite being proficient in Mu integration ., The data show that Mu insertions are repaired by the replication restart machinery and homologous recombination proteins ., A functional map of the Mu genome is shown in Figure 2A ., A ∼1 kb cat cassette encoding chloramphenicol ( Cm ) resistance was inserted into a non-essential region of the prophage genome ( see Materials and Methods ) ., Phage derived from this strain were used to infect the Keio mutant collection ( see Table 1 for strain information ) , which occupies forty-eight 96-well plates , and spotted on agar slabs containing chloramphenicol to select for Mu lysogens as described in Methods ., The control panel in Figure 2B shows results expected for known hosts where Mu integrates , but either does or does not replicate ., In our standard wild type host BU1384 where Mu replicates , ∼90% of the infected cells undergo lytic growth and lysis , and ∼10% of the survivors ( i . e . ∼1% of input cells ) are lysogens ., Mu fails to replicate in isogenic strains carrying either a himA or a clpX null mutant allele ., himA ( ihfA ) codes for one of the two subunits of the regulatory protein IHF , which is required for early Mu gene transcription 2 , 17 , and ClpX is essential for Mu replication 5 , 18 ., Both of these mutant strains support Mu integration 12 , 13 , 19 ., A larger number of CmR colonies are recovered in these strains compared to wild type because Mu does not undergo lytic development ., Similar differences in the recovery of CmR colonies were seen in the wild type Keio strain BW25113 and its isogenic himA and clpX derivatives ., In our screen for repair-defective mutants , we expected to identify mutant spots with either no CmR colonies or with fewer colonies than wild-type ., The majority of mutant strains behaved like wild type in this screen ., Known host mutants that do not support replication were easily identified ( Figure S1 , see plate #1 ) , but no new candidates with this phenotype were observed ., Several mutants displayed the phenotype of interest i . e . showed fewer or no colonies in the spots compared to wild type ( Figure S1 , see plate #1 and #9 ) ., The phenotype of these latter mutants was re-confirmed by infecting with Mu phage carrying a different antibiotic resistance marker ( Mu::Amp ) to ensure that the phenotype was independent of the antibiotic used for selection ., The final set of 30 mutants displaying this phenotype is arranged in four panels below the control panel in Figure 2B ., The mutants are classified broadly into genes known to affect DNA recombination/Repair , RNA-associated functions , ‘Other’ functions , and Mu receptor function ., A more detailed description of gene function is listed in Table S1 ., The poor yield of CmR colonies in the mutants shown in Figure 2B could be due to defects in Mu entry , integration , stable maintenance of lysogeny , or repair ., To distinguish between some of these possibilities a PCR assay was first employed to test for Mu integration ( Figure 3A ) ., Two primers were chosen to amplify covalent junctions between the left end of Mu DNA and an arbitrarily chosen target gene purH ., A PCR product is expected once the 3′ ends of Mu are joined to the target regardless of the fate of 5′ ends ( see Figure 1 ) ., PCR products of different lengths are expected since Mu integration is essentially random 20 , 21 ., Using this method , a control experiment first followed the time course of wild type as well as mutant Bam and Aam Mu phage infections in the wild type strain ., The particular Bam mutation used here ( Bam1066 ) is reported to be fairly proficient in integration but defective in replicative transposition of Mu 22 ., The Aam mutant ( Aam1093 ) is defective in integration 23 ., The integration patterns obtained during these infection experiments were consistent with the known transposition properties of these phages ( Figure 3A ) ., Wild type Mu was used to infect the 30 mutants obtained in the initial screen for repair-defective mutants ( Figure 2B ) ., Mutants grouped under Recombination-Repair , RNA and Other categories all showed similar levels as well as patterns of integration compared to the wild type strain ( Figure 3B ) ., Quantitative PCR with a subset of these mutants ( priA , recA ) validated the results with normal PCR ( Figure S2; we note that Southern blots used in earlier studies also showed similar levels of Mu integration in wild type and priA mutants 24 ) ., Thus , these mutants were not defective in either Mu entry or integration ., A majority of the mutants with defects in the LPS biosynthesis pathway , however , showed little or no integration ( Figure 3C ) ., This is likely due to a block in Mu entry , since the receptor for Mu is located within the LPS 25 , 26 ., To test if mutants that supported integration also supported Mu replication , cell lysis and phage production were monitored ., Growth of the strains with and without Mu infection is shown in Figure S3 ., The LPS mutants in Figure 2B all grew as well as wild type; only a representative mutant rfaF is shown in Figure S3A ., Neither this mutant , nor others in this category were susceptible to lysis by Mu infection ( Figure S3B ) , supporting the conclusion that this group of mutants is defective in Mu entry ., They were therefore not studied further ., The remaining mutants showed varying degrees of growth impairment compared to wild type ( Figure S3A ) ., With the exception of priA and dnaT , which are essential for Mu replication 5 , cell lysis and phage production were observed in all of the infected strains ( Figure S3B ) ., Thus , the majority of these mutants supported both Mu integration and replication ., Their defect in yielding stable lysogens could therefore be due to an inability to maintain lysogeny or defects in repair of the insertions ., Defects in maintenance of the prophage state or lysogeny might be discerned by examining Mu plaque morphologies on these mutants ., These would be expected to have a ‘clear’ rather than the ‘turbid’ phenotype observed for wild type Mu , which can be maintained in a lysogenic state ., dksA , hfq , rnt and rpsF gave turbid plaque morphologies somewhat similar to the wild type strain , dedD was apparently clear , while the remaining mutants had clear centers and clear edges with turbid rings in-between ( Figure S4 ) ., In the latter set of mutants with the mixed clear-turbid phenotype , it was difficult to ascertain whether the lysogeny-maintenance function might be affected ., To eliminate scoring mutants as repair-defective because they were unable to maintain the lysogenic state and were therefore going lytic , we re-screened the Keio library with a Mu::Cm variant defective in replication ., This phage carries the Bam1066 mutation , which allows integration but does not support replicative transposition ( see Figure 3A; 22 ) ., The same set of mutants was isolated in this screen as well ., In the spot test results shown in Figure S5 , it appears that some of the mutants have more CmR colonies than obtained with wild type phage ( see Figure 2B ) ., This is because a higher proportion of cells survive during infection with this phage due to absence of lytic growth ., Lysogen recovery was therefore quantified as described under Methods ( Figure 4A ) ., Among mutants in the Recombination-Repair category , priA and dnaT mutants were the most severely affected in lysogen recovery ( 0 . 04% ) , followed by recA ( 0 . 2% ) , recB ( 0 . 7% ) and recC ( 0 . 9% ) ., Among mutants in the RNA and Other category , with the exception of yfgL , dksA , hfq , rimK and lpd , the remainder had lysogen frequencies similar to or even better than wild type ., A surprising aspect of the data shown in Figure 4A is that lysogen recovery in the wild type was only ∼5% with MuBam phage , and that cell viability after infection was only ∼20% ( Figure S6A ) ., Similar low cell viability was observed even after infection with integration-defective MuAam phage ( Figure 3A and Figure S6A ) , which gave no CmR colonies ., To test if this was due to expression of the cell killing function kil or to other function ( s ) specified by the unknown orfs in the SE ( semi-essential ) region 27 , which is transcribed as part of a long early transcript that includes the A and B genes 2 ( see Figure 2A ) , we deleted the SE region in the MuBam phage ( see Methods ) ., Indeed , infection with MuBam1066ΔSE::Cm phage improved both lysogen recovery and cell viability in the wild type to 100% ( Figure 4B and Figure S6B , respectively ) ., Under these conditions , all the mutants in the Recombination-Repair category still remained impaired ( <15% of wild type ) for lysogen recovery ., In the RNA/Other category , hfq , lpd and lipA were also still substantially impaired ( 18–25% of wild type ) ., Since hfq shows wild type plaque morphology ( Figure S4 ) and since there is no obvious relationship of the known functions of these three genes to DNA repair , we will not consider them further here ., We conclude that a majority of the E . coli genes required for recovery of stable Mu insertions provide functions that apparently allow host survival in the presence of lethal phage functions specified by the SE region of Mu ., The group of five genes that remain defective - priA , dnaT , recA , recB and recC – is significant in that this group is known to participate in recombinational repair ., The isolation of this group of genes must be related to the repair of Mu insertions and not to repair of random double strand breaks generated upon Mu infection , because ( 1 ) they are dependent on Mu integration ( i . e . infection with MuAam1093 phage does not significantly affect the viability of the priA and recA hosts as compared to wild type; Figure S6A ) , and ( 2 ) Mu-induced mutations are known to be tightly linked to Mu i . e . they are not random 28 ., PriA and DnaT play a central role in the repair of nicks and gaps created by DNA damaging agents in E . coli by promoting replication restart after fork collapse , either with or without the involvement of recombination 29 ., There are multiple pathways for replication restart that require PriA , PriB , PriC , DnaT and Rep 29 ., These proteins identify the correct substrate , process it if necessary , and then aid DnaC in loading the replicative helicase DnaB during pre-primosome formation ., PriA and DnaT are required for the two main pathways of ‘Restart’ where PriB and PriC have redundant roles ., Thus priA and dnaT null mutants have extreme phenotypes whereas priB and priC null mutants have none ., dnaC809 , 820 is a priC/rep-independent suppressor that restores all known phenotypes of priA and dnaT null mutants 30 ., During the lytic cycle of Mu growth , PriA restarts Mu replication without the involvement of homologous recombination ( 24 , 31 and Figure S3B ) ., The data reported in Figure 2 , Figure 3 , and Figure 4 in this study show that PriA and DnaT are also required during the non-replicative event , along with a requirement for homologous recombination proteins ., To confirm the phenotype of priA , dnaT , and the rec genes and to dissect the role of PriA further , we tested these and several different mutant alleles of these genes in a different strain background ., The priA , dnaT , recA , recB ( and recBCD ) mutants all showed defects in Mu lysogen recovery in this strain background as well ( Figure 5A ) ., priA and dnaT mutants show poor growth ( Figure S3A and 32 , 33 , 34 ) and many cells in the population have high levels of SOS expression 35 ., SOS genes are normally kept silent by the repressor LexA , and activated only when LexA is cleaved by RecA in response to DNA damage 36 ., SOS induction can be prevented by removing recA or by introducing a non-cleavable lexA3 allele 37 ., To test if SOS expression is responsible for the low recovery of Mu lysogens , we tested priA lexA3 and priA recA double mutants; both mutants remained defective ( Figure 5B ) ., A lexA3 mutant alone supported efficient recovery of Mu insertions , showing additionally that the SOS response is not required , but that the recombination function of RecA is needed ., We note that recA1 , a recombination-defective missense allele of recA , was not seen to affect recovery of Mu insertions in Salmonella 38 , 39 ., This allele can bind ssDNA in vitro 40 , and perhaps has residual activity in vivo that allows it to function in Mu repair ., We also note that several genes in the Keio collection were recently reported to be partially duplicated 41 ., Of these , priB and polA are of interest to this study ., These gene deletions as well as priC were therefore re-tested in the same strain background as the priA alleles ., They were found to not affect Mu recovery ( Figure 5B ) ., PriA has at least four types of activities: ATPase , helicase , the ability to load the replisome , and the ability to interact with other proteins ., PriA300 ( K230R ) inactivates the ATPase and helicase activities , yet primosome assembly can occur both in vivo and in vitro 42 , 43 ., PriA301 ( C479Y ) mutates a residue in the cysteine-rich region of PriA thought to be important for protein-protein interactions and helicase activity 44 ., Like priA300 , priA301 maintains wild-type growth and recombination proficiency 45 , 46 ., Lack of the helicase activity of PriA has been reported to impair Mu replication both in vivo and in vitro 31 ., Using the helicase-defective strains priA300 and priA301 , we observed that the helicase and protein-protein interaction activities of PriA are largely dispensable ( Figure 5C ) , indicating that it is the primosome activity of PriA that is essential for recovery of Mu insertions ., This is further supported by the observation that combining priA and dnaT null mutations with dnaC809 , 820 restores the ability of strains to recover lysogens ( Figure 5C ) ., Both in vivo and in vitro experiments have suggested that mutant DnaC proteins suppress the absence of PriA/DnaT complex by bypassing its role in helping DnaC to load DnaB/PolIII directly onto a recombinational intermediate 30 , 47 ., To confirm that all of these data point to a critical role for replication restart in repair of Mu insertions , we sequenced fifteen independent insertions which were recovered at a low frequency in the priA mutant ( see Materials and Methods ) ( Figure 6 ) ., Of these , five insertions had rearranged the Mu-host junctions in various ways , and their precise location could not be determined ., Two insertions had symmetrical additions ( at both ends ) of a nucleotide not found in the wild type host , likely due to repair by an error-prone polymerase , and one of these strains had two copies of Mu ., Eight insertions had normal Mu-host junctions ., We note that the sequencing strategy included cloning of CmR Mu DNA fragments , favoring recovery of R end fragments that had not been deleted or rearranged , and therefore underestimating the fraction of incorrectly repaired insertions ., Overall , these results show that in the absence of PriA , Mu insertions are repaired inefficiently and often incorrectly by alternate pathways ., Thus , PriA is indeed required for normal repair of Mu insertions ., There are three pathways for replication restart in E . coli: PriA–PriB , PriA–PriC , and PriC–Rep , which differ in their recognition of stalled forked structures 29 ., PriA plays an essential role in initiation of replication on the forked DNA intermediates generated during the lytic phase of Mu growth , using either the PriA–PriB or PriA–PriC pathway , in addition to the proteins that are required for E . coli chromosomal replication 24 , 31 , 53 , 54 ., During Mu transposition , the transition from strand transfer to DNA replication can be divided into a number of discrete steps 3 , 5 ., MuA initially remains tightly bound to the Mu fork as a multi-subunit complex called transpososome ., In a highly choreographed series of steps , host proteins dislodge this transpososome and assemble a replisome ., In the first step of this transition , ClpX alters MuA subunit interactions to weaken interaction of the transpososome with DNA 55 , 56 , 57 ., Next , as yet unidentified cellular factors called Mu Replication Factor α2 ( MRF α2 ) displace the transpososome and exchange it with the translation initiation factor IF2-2 to produce a pre-replisome 58 ., Finally , the helicase activity of PriA is required to displace IF2-2 , remodeling the template to permit replisome assembly , which includes DnaT , DnaB , DnaC and the DNA polymerase III holoenzyme 5 ., PriA has distinct replisome assembly and 3′ to 5′ helicase activities 29 ., Helicase-defective PriA supports little or no Mu replication in vitro , and shows a partial defect in Mu replication in vivo 31 ., These data indicate that PriAs replisome assembly activity is essential for initiation of Mu DNA replication and that the helicase activity also promotes this process ., PriA is thought to bind to the lagging strand template at the fork and unwind it in a 3′ to 5′ direction , promoting loading of DnaB , thus coupling its replisome assembly and helicase activities ., The surprising requirement of PriA and DnaT in the non-replicative pathway of Mu transposition as reported in this study , suggests strongly that the 5 bp gaps generated upon Mu insertion are repaired by the replication restart machinery ., This shared requirement for the PriA primosome in both pathways might imply that the PriA loading steps after strand transfer are similar in both ., What apparently distinguishes the two pathways is non-requirement of the helicase activity of PriA , and requirement for homologous recombination proteins ., We discuss two alternate models for recombinational gap repair below ., Nicks and gaps in DNA are normally repaired when their encounter with a traveling replication fork converts them into a double strand break , collapsing the fork 36 ., The broken end serves as an entry point for RecBCD , generating single strands for RecA binding , followed by invasion of the intact sister chromosome , thus reconstituting a forked structure for restarting replication via the PriA primosome 59 , 60 ., In such a scenario for Mu repair , an oriC-initated fork will cause a double strand break when , arriving at the site of a Mu insertion , it encounters the flanking gap ( Figure 7A ) ., The double-strand break will be on the chromosomal DNA flanking the Mu insertion , which is expected to be processed by RecBCD , followed by restoration of the fork by recombination , and restart of replication by the primosome ., Two considerations make this scenario unappealing ., First , Mu does not insert near replication forks 61 , so the unrepaired intermediate would be potentially vulnerable to degradation while it waits for the oriC-initiated fork to arrive ., Second , the passing fork would encounter only one of the two gaps at each Mu end that need repair , so the entire Mu would have to be replicated , generating a second double strand break at the distal Mu end , reiterating RecA-mediated invasion and primosome assembly before repair of the second gap can be completed ., A parsimonious alternative model takes advantage of the PriA replisome already present at the forked strand transfer joints at both Mu ends , recruited there in the normal course of transpososome disassembly ( see Figure 1 ) ., In this model , the initial steps of PriA recruitment and replication are common to both the repair and replication pathways ( Figure 7B ) ., The pathways differ in the flap cleavage step , which ensues concomitant with replication restart , leaving double-strand breaks on the Mu lagging strand ., These breaks allow RecBCD entry , creating single-stranded 5′ Mu ends on which RecA polymerizes 62 ., Although 3′ end strand invasion is generally preferred with purified RecA , 5′ ends can be used for strand exchange in vitro 63 , and in vivo recombination data also fit models that invoke 5′ strand invasion 64 ., The Holliday junction so created can then be resolved by Ruv proteins or endonucleases ., This model reverses the steps normally associated with recombinational repair , with replication preceding recombination ., According to this model , there will be limited replication near the two Mu ends in this largely non-replicative event ., What signals flap cleavage in one pathway and not in the other ?, We speculate that the MuN protein , which normally protects the ends of infecting Mu DNA from degradation , dissociates from the ends , perhaps upon interaction with the transpososome assembled on the strand transfer complex ., This allows RecBC to enter and peel away the 3′ strand of the flap , engaging and activating MuANuc on the 5′ strand ., This is the first report of specific host processes involved in repair of transposon insertions in bacteria ., We find that the PriA primosome and homologous recombination proteins , which are essential for repair of double-strand breaks in E . coli , play a critical role in the repair of Mu insertions ., We favor a model for recombinational repair in which PriA restart of Mu replication is followed by RecA-mediated resolution of double-strand breaks on the Mu lagging strands created by the flap endonuclease activity of the transposase ., Given that the predominant route taken by Mu upon infection is to enter lytic growth , it is plausible that Mu first co-opted the PriA system for replication , and later used it for repair ., It will be interesting to see whether other transposons use these same processes for repair of their insertions ., All strains used in this work are derivatives of E . coli K-12 and are listed in Table 1 13 , 19 , 22 , 23 , 34 , 35 , 65 , 66 ., The Keio Collection ( single-gene knockout library of 3 , 985 nonessential genes in E . coli ) was obtained from the National BioResource Project , Japan ., The wild type strain in this collection is BW25113 ., E . coli Mu lysogen strains BU1717 or MH3491 were used to construct strains SJ17 – SJ19 ( Table 1 ) , where a ∼1 kb cat cassette was inserted downstream of the invertible G-segment on the Mu genome at nt 35 , 040 , before gin , by the method of Datsenko and Wanner 67 ., The SE deletion was similarly constructed; it removes nt 4 , 319–7 , 954 from the Mu genome , substituting the cat cassette in its place ., All Mu phages used in this study carry the temperature-sensitive ts62 allele of the lysogenic repressor gene c ., Primers used in this study are listed in Table S2 ., Cultures from the Keio collection stocked in 96-well plates were inoculated into new sterilized 96-well plates with 0 . 2 ml of Luria broth ( LB ) by using the 12-multichannel pipette ( Biohit ) ., They were incubated at 37°C overnight without shaking ., 4 µl of saturated overnight cultures were transferred to 0 . 2 ml of fresh LB media supplemented with 2 . 5 mM CaCl2 and 5 mM MgSO4 in 96-well plates and incubated at 37°C until OD600 reached around 0 . 5 , measured directly in the plates by DTX880 microplate reader ( Beckman ) ., Mu phage was added to the cultures at a multiplicity of infection ( moi ) of 5 , mixed briefly , and incubated at 30°C for 1 hr . 4 µl of infected cultures were spotted on slab agar plates having dimensions similar to the 96-well plates and containing 25 µg/ml chloramphenicol; plates were incubated overnight at 30°C ., 50 µl overnight cultures were transferred to 5 ml of fresh LB media supplemented with 2 . 5 mM CaCl2 and 5 mM MgSO4 and grown to 0 . 5 at an OD600 ., Phage were added to the cultures at 5 moi and incubated at 30°C for 30 min ., Infected cells were harvested and the total DNA were isolated by Wizard Genomic DNA purification kit ( Promega ) ., PCR was conducted with 50 ng DNA as a template , 10 pmol primers , 1× Go Taq master mix ( Promega ) , and distilled water up to 50 µl ., Primers were designed to anneal to the left end of Mu DNA and the purH gene of E . coli ., PCR conditions were: 94°C for 2 min , 30 cycles of - 94°C for 30 sec , 50°C for 30 sec , 72°C for 2 min 30 sec - and a final extension at 72°C for 2 min ., PCR amplification primers used in this study are listed in Table S2 ., The reaction products were electrophoresed on 1% agarose gels and visualized by staining with ethidium bromide ., This method measures DNA amounts based on the fluorescence signal from SYBR-bound DNA ., PCR reactions were conducted with the same templates and primers as used for normal PCR , with the additional inclusion of 1× Power SYBR Green PCR Master Mix ( Applied Biosystems ) , and distilled water up to 25 µl ., The PCR program in the 7900HT sequence detector ( Applied Biosystems ) was as follows: 95°C for 10 min , followed by several cycles of - 95°C for 30 sec , 50°C for 30 sec , 72°C for 2 min 30 sec ., Cumulative fluorescence was measured at the beginning of the exponential phase of the PCR reaction to determine the fractional cycle number ( CT ) ., The level of integrated Mu DNA was normalized to a chromosomal locus dnaC , amplified with appropriate primers listed in Table S2 ., 100 µl of saturated overnight cultures were transferred to 10 ml of fresh LB media and incubated at 37°C until OD600 reached around 0 . 5 for all cultures ., From then on , growth was monitored by measuring OD600 at various times for 2 hr ., A similar procedure was followed for obtaining lytic growth curves , except that the LB media was supplemented with 2 . 5 mM CaCl2 and 5 mM MgSO4 ., At OD600 of around 0 . 5 , Mu phage was added at 5 moi , mixed briefly , and incubated at 37°C for 3 hr until most cultures were completely lysed ., In all cases where priA2::kan , dnaT822 ( without dnaC mutations ) or polA::kan strains were used , these were grown overnight in minimal media , followed by dilution into fresh LB media , and then allowed to grow into log phase before infection with the different Mu phages ., These were prepared by induction of the prophage strains by thermal inactivation of the temperature-sensitive ( ts ) phage repressor c , and concentrated by CsCl gradient centrifugation as described 12 ., For strains BU1717 ( MuBam1066 ) , SJ17 ( Mu::Cm ( Bam1066 ) ) and SJ18 ( MuBam1066ΔSE::Cm ) , the prophages were induced in the presence of pJG4 ( c-myc MuB expressed from pET28 ( a ) without IPTG induction ) to supplement MuB protein ., Typical phage titers after concentration were ∼1011 pfu ( plaque forming units ) for wild type Mu , and ∼1010 pfu for the Bam or BamΔSE phage ., Phage titers for wild type Mu with and without the cat insertion were similar , showing that the insertion did not affect phage yields ., Cultures were infected with Mu::Cm ( Bam1066 ) , MuBam1066ΔSE::Cm or Mu::Cm ( Aam1093 ) phage as described under ‘PCR-based assay for Mu integration’ ., Before and after infection , appropriate dilutions of cells in LB media were spread onto agar plates with or without 25 µg/ml chloramphenicol to obtain cell counts for input cells , survivors after infection , and lysogens ., Plates were incubated at 30°C overnight , and colonies were counted the next day ., Lysogenization efficiency was calculated as CmR cells/input cells×100 , and survival efficiency was calculated as survivors ( on non-antibiotic plate ) /input cells×100 ., priA lysogens were selected as CmR colonies after infection with Mu::Cm ( Bam1066 ) phage ., After overnight culture into LB media , chromosomal DNA was isolated by Wizard Genomic DNA purification kit and digested by restriction enzyme BamHI and PstI ., Digested DNA fragments were purified and ligated with similarly digested pUC19 plasmid ., CmR transformants were isolated and digested by BamHI and PstI to ascertain that the insert size was larger than 4 kb , so that it included DNA flanking the insertion ., R1 primer ( Table S2 ) was annealed to Mu DNA right end to obtain sequence of the flanking DNA ., Based on this sequence , appropriate primers were used to PCR-amplify DNA flanking the left end of the insertion using the L1 primer ., DNA sequencing was performed at our core sequencing facility ., 10 µl of an appropriate dilution of phage suspension were mixed with 100 µl of host cells grown to 0 . 5–0 . 6 at OD600 in LB including 2 . 5 mM CaCl2 and 5 mM MgSO4 ., The mixture was added to 3 ml of 0 . 3% molten soft agar at 42°C , and poured on top of an LB agar plate containing 2 . 5 mM CaCl2 and 5 mM MgSO4 ., Plates were incubated overnight at 37°C . | Introduction, Results, Discussion, Materials and Methods | Mu is both a transposable element and a temperate bacteriophage ., During lytic growth , it amplifies its genome by replicative transposition ., During infection , it integrates into the Escherichia coli chromosome through a mechanism not requiring extensive DNA replication ., In the latter pathway , the transposition intermediate is repaired by transposase-mediated resecting of the 5′ flaps attached to the ends of the incoming Mu genome , followed by filling the remaining 5 bp gaps at each end of the Mu insertion ., It is widely assumed that the gaps are repaired by a gap-filling host polymerase ., Using the E . coli Keio Collection to screen for mutants defective in recovery of stable Mu insertions , we show in this study that the gaps are repaired by the machinery responsible for the repair of double-strand breaks in E . coli—the replication restart proteins PriA-DnaT and homologous recombination proteins RecABC ., We discuss alternate models for recombinational repair of the Mu gaps . | Transposon activity shapes genome structure and evolution ., The movement of these elements generates target site duplications as a result of staggered cuts in the target made initially by the transposase ., For replicative transposons , the single-stranded gaps generated after the initial strand transfer event are filled by target-primed replication ., However , the majority of known transposable elements transpose by a non-replicative mechanism ., Despite a wealth of information available for the mechanism of transposase action , little is known about how the cell repairs gaps left in the wake of transposition of these majority elements ., Phage Mu is unique in using both replicative and non-replicative modes of transposition ., Our study finds that during its non-replicative pathway , the gaps created by Mu insertion are repaired by the primary machinery for double-strand break repair in E . coli , not by gap-filling polymerases as previously thought ., This first report of specific host processes involved in repair of transposon insertions in bacteria is likely to have a broad significance , given also that double-strand break repair pathways have been implicated in repair of the retroviral and Line retroelement insertions . | biology | null |
journal.pbio.1001135 | 2,011 | Video-Game Play Induces Plasticity in the Visual System of Adults with Amblyopia | The most frequent cause of permanent visual loss in childhood is amblyopia ( “lazy eye” ) 1 , 2 , a developmental disorder associated with early abnormal visual experience that disrupts neuronal circuitry in the visual cortex and results in abnormal spatial vision ., It is generally believed that adult amblyopia is irreversible beyond the sensitive period of brain development ., However , new studies , both in humans 3–12 and in rodents 13–15 , suggest that the mature amblyopic brain retains a substantial degree of plasticity ., In particular , human adults with long-standing amblyopia show substantial improvements in performing a visual task , following perceptual learning ( extended practice ) of the task ., Playing video games results in enhancement of a broad range of visual tasks in adults with normal vision , including light sensitivity 16 , contrast sensitivity 17 , visual crowding 18 , and visual attention 19 ., However , while it is now clear that video-game play can strengthen some aspects of normal vision , it is not clear whether video-game play can induce functional plasticity in the mature visual system following a prolonged period of abnormal development ., Moreover , while video-game play improves the spatial resolution of attention in normal participants , it does not improve visual acuity ( with isolated targets ) ., Since reduced visual acuity is the sine qua non of amblyopia , it is crucial that video-game play can improve visual acuity if it is to be a useful tool for visual rehabilitation in patients with reduced spatial vision ., In the present study , we aimed to assess with a small pilot group whether playing video games with an amblyopic eye can induce cortical plasticity and improve spatial vision in adults with amblyopia , well beyond the “sensitive period” of brain development ., We hypothesized that the intense sensory-motor interactions while immersed in video-game play might push brain functions to the limit , enabling the amblyopic visual system to learn , on the fly , to recalibrate and adjust , providing the basis for functional plasticity ., Moreover , game playing requires the allocation of spatial attention , detection , and localization of low contrast , fast moving targets , and aiming ( in first-person shooter games ) ., Thus , we speculated that video games may include several essential elements for active vision training to boost visual performance , and thus could potentially be useful in improving amblyopic vision ., We tested a range of visual functions to examine the neural alternations , if any , following video-game play in a small group of adults ( Figure 1 ) ., These visual functions , ranging from low-level to high-level vision , included visual acuity ( letter acuity ) , positional acuity ( Vernier acuity ) , visual counting ( spatial attention ) , and stereoacuity ( 3-D binocular vision ) ., In order to understand the neural mechanisms that underlie the video-game experience induced visual plasticity , we measured and modeled a positional acuity task in noise ., While action video games are reported to be useful in enhancing visual function in normal humans , non-action video games are not effective 19 ., Playing action video games may not be ideal for patients with amblyopia , particularly children ., Therefore , in another set of pilot experiments , we also examined whether non-action video games may be effective for recovering amblyopic visual functions ., Our participants played video games for 40 h with their fellow eye patched ., One might argue that the visual improvements , if any , might have resulted from the eye patching alone ., To address this point , we used a cross-over treatment design in which a group of amblyopes first underwent occlusion therapy , i . e . patching the fellow sound eye , for a period of time before the video-game phase ., With this experimental design , we can compare the efficacy of the two treatment approaches ( passive patching and video-game playing ) ., Our study had several limitations: small sample size , lack of randomization , and differences in numbers between groups ., A large-scale randomized clinical study is needed to confirm the therapeutic value of video-game treatment in clinical situations ., Nonetheless , taken as a pilot study , this work suggests that video-game play may help guide future treatment of amblyopia ., To evaluate how video-game play alters amblyopic vision , we monitored the changes , if any , in visual acuity in 10 adults with amblyopia while they played a first-person shooter game—Medal of Honor: Pacific Assault ( MOH ) —using their amblyopic eye , with the fellow sound eye patched with a black eye patch ., Visual acuity ( VA ) is a standard clinical procedure to quantify spatial vision by determining the smallest letter on a chart that can be identified at a given viewing distance ., In amblyopia , vision is often substantially poorer when the target letter is presented with surrounding letters than when it is presented alone , a phenomenon known as crowding 20 ., Therefore we measured both crowded line-letter acuity and uncrowded single-letter acuity so as to provide a comprehensive evaluation of visual acuity ., Surprisingly , playing video games rapidly reversed their amblyopia ., After 40 h of video-game play ( 2 h/d ) , acuity improved , on average , by 1 . 6 and 1 . 4 lines on a LogMAR letter chart for crowded letters and single letters , respectively ( Figure 2a , top panels ) , representing 31 . 2%±3 . 1% ( crowded: t\u200a=\u200a10 . 154 , p<0 . 0001 ) and 27 . 2%±3 . 2% ( uncrowded: t\u200a=\u200a8 . 598 , p<0 . 0001 ) improvements in minimum angle of resolution ( MAR—bottom panels ) ., Two mild amblyopes ( SA2 & SA4 ) completely “normalized” according to a criterion of 20/20 ( LogMAR\u200a=\u200a0 , dotted line ) ., It might be argued that the improvements could be due to learning the letter charts ., Therefore , instead of taking measurements every 10 h , we tested observer SS1s acuity only before and after the video-game intervention , and similar to what we observed in other observers , his acuity improved substantially ( ≈2 . 5 letter-lines or ≈44% for both measurements ) ., While it has been clearly demonstrated that playing action video games improves a broad array of visual functions in adults with normal vision , non-action games are not effective 17 , 18 ., For example , playing action video games resulted in enhanced crowded resolution acuity in normal vision , while playing a non-action video game did not ., However , action games may not be ideal for patients with amblyopia , particularly younger patients ., Therefore , in the next experiment , we asked another three amblyopic patients to play a non-action video game—SimCity Societies ( SIM ) ., Interestingly , we found that similar to the action game group , all three non-action game players showed enhanced vision ( Figure 2b , phase 1: 0 to 40th h ) , and one , a mild amblyope ( SA6 ) normalized to ≈20/16 ., On average , this group was able to read 1 . 5 more letter-lines ( 28 . 4% improvement ) for crowded-letter acuity and 0 . 8 more lines ( 15 . 1% improvement ) for single-letter acuity ., These findings suggest that non-action games share useful properties for enhancing amblyopic vision ., To determine the limits of plasticity , the three players who participated in the SimCity experiments were then asked to play MOH for another 40 h ( phase 2: 40th to 80th h ) ., Additional improvements of about one letter-line ( SB2 & SA5 , crowded: 18% ) were observed ., Note that SA6s amblyopia was completely normalized at the end of phase I and no further significant improvement was observed ., Since our participants played video games with the fellow eye patched , the vision enhancement we observed could have been the result of wearing an eye patch alone ., Thus , in a control experiment , another group ( OT ) of seven amblyopic adults wore a patch , but instead of playing video games they were required to engage in other visually demanding activities , such as watching television , reading books , knitting , and surfing the Internet , using the amblyopic eye ., After 20 h , however , no significant change in acuity was observed ( Figure 2c , phase 1: 0–20th h ) ; the dashed line in the bottom panels shows the mean data ( OT20: crowded: mean improvement\u200a=\u200a0 . 4%±3 . 0% , t\u200a=\u200a0 . 1317 , p\u200a=\u200a0 . 8995; uncrowded: mean improvement\u200a=\u200a−3 . 7±3 . 2% , t\u200a=\u200a1 . 136 , p\u200a=\u200a0 . 2991 ) ., In contrast , for the same amount of time , the video-game group ( n\u200a=\u200a9 ) showed a marked improvement in acuity of ≈20% ( Figure 2a , MOH20>OT20: crowded: t\u200a=\u200a4 . 337 , p\u200a=\u200a0 . 0007; uncrowded: t\u200a=\u200a3 . 74 , p\u200a=\u200a0 . 0022 ) ., Five of the seven participants who completed the patching experiment continued to the video-game phase for another 40 h ., In this phase of the experiment , we used both action ( all except SC1 ) and non-action ( SC1 ) games ., Although none of the five showed any significant change in acuity in the patching phase , all improved substantially in the video-game phase ( OT-VG20: ≈1 . 7 letter-lines , ≈29% improvement in both measurements; OT20<OT-VG20: crowded: paired t\u200a=\u200a5 . 712 , p\u200a=\u200a0 . 0046; uncrowded: paired t\u200a=\u200a2 . 785 , p\u200a=\u200a0 . 0495 ) ., From this small-scale “cross-over” experimental design , we can conclude that it is the video-game experience , and not simply the patching , that enhances amblyopic vision ., Figure 2d summarizes all the acuity data from the above experiments ., The mean improvement in visual resolution across all 18 participants who completed the video-game training from the three experiments was ≈30% ( crowded acuity: 1 . 8 letter-lines , 33 . 4%±2 . 4% and isolated acuity: 1 . 5 letter-lines 27 . 4%±3 . 5% ) ., The effect sizes ( Cohens d value ) at the 20th h were 3 . 03 and 1 . 33 for crowded acuity and isolated acuity , respectively ., The recovery of crowded acuity was slightly faster than uncrowded acuity ., An exponential fit y\u200a=\u200ayo+a ( 1−e−bx ) to the data revealed time constants ( b ) of 0 . 064 and 0 . 054 h−1 for crowded acuity and uncrowded acuity , respectively ., It is worthwhile noting that the recovery rate we observed here in adults is ≈5-fold faster when compared with the conventional occlusion therapy in children ., It would take >200 h to obtain comparable treatment effects in children ( ≈0 . 1 logMAR unit/120 h ) 21 , and it would be reasonable to expect a much longer treatment course for adults 22 ., There was no significant correlation between the amount of acuity improvement and the baseline acuity ( Figure 2d bottom left ) ., The mean crowding index , crowded acuity ( MAR ) / uncrowded acuity ( MAR ) , was slightly , but not significantly , reduced ( by 5 . 9%±4 . 3% ) , indicating that video-game play improved crowded acuity slightly more than uncrowded acuity ( Figure 2d bottom right ) ., While visual acuity represents one important limit to spatial vision , positional acuity , which represents the ability to localize visual objects , is another important aspect of spatial vision ., While positional acuity is remarkably acute in normal vision ( often referred to as hyperacuity ) , it is often severely impaired in amblyopia ., We found that positional acuity ( the ability to detect a misalignment between the two line segments—Figure 3a ) improved significantly following video-game play ( on average 16 . 0%±4 . 0%; n\u200a=\u200a16 MOH40: 12+SIM40: 4—Figure 3b , black solid line , zero external noise: t\u200a=\u200a3 . 963 , p\u200a=\u200a0 . 0012; non-amblyopic eye: 1 . 0%±10% , t\u200a=\u200a0 . 1057 , p\u200a=\u200a0 . 9179 ns ) ., To understand the neural mechanisms underlying this improvement , we introduced positional noise 23 to mimic the spatial distortions ( internal spatial noise ) existing in the visual system and applied a positional averaging model to the data ( see Materials and Methods ) ., Figure 3c shows that the ability to extract and process information from the visual stimuli for positional averaging was actually boosted by 33 . 1% , with mean sampling efficiency improved from 6 . 8% to 9 . 0% ., Some observers also showed reduced spatial distortion ( on average internal noise decreased by 7%—Figure 3d , from 0 . 185 λ to 0 . 172 λ ) , indicating that the distorted retinotopic cortical mappings were recalibrated and less distorted ., Fig 3e summarizes the different neural mechanisms ( SB2: sampling efficiency enhancement; SA5: spatial distortion reduction; SS3: combination of both ) that underlie the improvement in positional acuity ., Video-game play also appears to increase visual attention in amblyopia ., We used a visual counting task to determine how many visual locations the brain can direct attention to in a very brief time period , 200 ms ( Figure 4a ) ., Previous work has shown that some amblyopes show severe deficits in visual counting 24 and that action game play can enhance counting in normal vision 19 ., In general , participants who initially showed the largest deficits in counting performance also showed the most improvement ( Figure 4c ) ., A subgroup of five participants ( symbols surrounded by dotted circles in Figure 4b ) showed significant undercounting ( Figure 4d , blue line ) ., For example , when 10 dots were displayed , the mean number of dots reported was 7 ( undercounted by 3 dots or 30% ) ., Undercounting is thought to reflect high-level neural deficits in amblyopia 24 ., Following video-game play , for the range of 7–10 dots , undercounting decreased significantly by 8 . 4% ( from 25 . 3% to 16 . 9%—Figure 4d , two-way RM ANOVA: F\u200a=\u200a33 . 022 , p\u200a=\u200a0 . 005; non-amblyopic eye: pre 5 . 6%→post 5 . 0% , two-way RM ANOVA: F\u200a=\u200a0 . 609 , p\u200a=\u200a0 . 492 ns ) ., The mean counting threshold ( the number of dots that can be reliably counted ) increased significantly by 37% , from 3 . 3±0 . 3 to 4 . 4±0 . 4 dots ( Figure 4e , paired t\u200a=\u200a4 . 508 , p\u200a=\u200a0 . 0108; non-amblyopic eye: pre 7 . 7±0 . 3 dots→post 8 . 0±0 . 3 dots , paired t\u200a=\u200a1 . 161 , p\u200a=\u200a0 . 3102 ns ) and the mean response latency decreased by 16 . 5% ( Figure 4f , N\u200a=\u200a1–10 ) , though not significantly ( two-way RM ANOVA: F\u200a=\u200a0 . 839 , p\u200a=\u200a0 . 424 ) ., In short , video-game play increases the number of items the amblyopic brain can direct attention to simultaneously , reduces undercounting deficits , and increases the processing speed of visual counting ., Amblyopia is associated with abnormal binocular vision and reduced or absent stereopsis ( binocular depth perception or 3-D ) ., With improved monocular vision following video-game play , for some amblyopes binocular vision also recovered to a substantial extent ., Five of the six anisometropic amblyopes ( with straight eyes ) were tested for stereopsis following the training ., All five showed improved stereopsis ( Figure 5a , n\u200a=\u200a5 MOH40: 3+SIM40: 2 ) ., Mean improvement in stereoacuity was 53 . 6%±8 . 4% ( Figure 5b , t\u200a=\u200a6 . 410 , p\u200a=\u200a0 . 003 ) , noting that SA6 failed the stereo test and had no recordable stereopsis in the baseline session ., Three participants ( SA2 , SA3 , and SA4 ) fully regained normal stereoacuity ( 20 arc sec ) as measured by this test , and were basically “cured” in this aspect of vision ., Here we provide evidence from a pilot study of a small number of people that video-game play can induce a substantial degree of visual plasticity in adults with amblyopia ., After a brief period of video-game play , a wide range of spatial vision functions improve very rapidly and substantially , reflecting normalization of both low-level ( visual acuity , positional acuity ) and high-level ( spatial attention , stereoacuity ) visual processing ., Importantly , we provide preliminary characterization of the time course , limits , and underlying mechanisms of video-game experience-dependent cortical plasticity ., The findings of our “cross-over control” experiment show that the treatment effects cannot be simply explained by eye patching , suggesting that it is indeed the video-game experience which improves amblyopic vision ., The visual plasticity stimulated through video-game training has been well documented in the “normal” visual system , however the neural mechanisms are not yet clear ., Using positional noise , we are able to reveal the underlying mechanisms ., As we previously reported , repeatedly practicing a Vernier task in positional noise , with response feedback , can improve sampling efficiency and re-calibrate the distorted retinal topographical mappings of the amblyopic visual field 11 , 17 ., Here we show that video-game play also results in a substantial increase in the ability to extract visual information ( increased sampling efficiency ) , without specific direct training , and we found that spatial distortion ( or internal positional noise ) can also be reduced to a certain extent through video-game play ., Our findings also provide insights into which levels of processing in visual cortex can be modified ., Counting deficits in amblyopes are thought to reflect a higher level limitation in the number of features ( and missing features ) the amblyopic visual system can individuate 24 ., We speculate that the reduction in undercounting deficits in our amblyopic participants may represent the normalization of these higher level cortical areas ., Recent work suggests that the ability to apprehend numbers may reflect a primary sensory attribute 25 , possibly reflecting the responses of neurons in parietal cortex that are tuned to number ., From this perspective , the response characteristics of these affected numerosity processing neurons might be modifiable with video-games experience ., While it is possible that low-level factors such as crowding 26 may result in improved counting in amblyopes , we can safely exclude this cause since our observers did not show any significant recovery in crowding ., Our regression analysis suggests that changes in crowded acuity account for 3% of the variance in counting threshold and changes in isolated acuity account for 77% ., Perhaps most importantly , we show that playing video games can indeed improve visual acuity and sharpen amblyopic vision ., Note that visual acuity is the gold standard for examining spatial vision in clinical situations ., To our knowledge , our work is the first to report that uncrowded visual acuity can be improved through video-game training ., Green and Bavelier 18 reported that 30 h of video-game play did not result in improved visual acuity in normal adults , perhaps because there is little room for improvement in the normal visual system , or because 30 h is simply not long enough to improve a function as fundamental as normal visual acuity ., Here we find that video-game play , both action and non-action , can result in a substantial improvement of amblyopic visual acuity ., This is especially important because reduced visual acuity is the sin qua non of amblyopia ., Playing a non-action game for 30 h has been found to be ineffective in enhancing attentional performance in participants with normal vision 19 ., However , our results suggest that not only action but also non-action video games might be effective in improving amblyopic spatial vision ., Although non-action games do not impose the same intense pressure on the player to respond to sudden pop-up targets from somewhere in the visual field , and to track fast moving objects , they do require the player to pay attention to fine and small spatial details and to different visual features in the visual scene—which may be a very demanding visual task for someone with reduced vision ., In fact we noted that during game play , some deep amblyopes initially required more time than normal participants and had to get closer to the screen in order to identify targets or read instructions ., In some sense , this is essentially similar to training spatial resolution 27 ., A long period of sustained attention in seeing fine visual details might play an important role in triggering neural plasticity ., It is worth noting that we had fewer participants , altogether four ( three from Group 2 and an extra one from the cross-over group SC1 ) , for the non-action video game ., We recognize that the treatment effects could vary from individual to individual ., A much larger sample size is necessary for future studies to investigate which type , action or non-action , is more effective in treating amblyopia ., Perceptual learning has shown to be useful in improving amblyopic vision 28 ., It is worthwhile noting that the visual recovery , e . g . visual acuity and positional acuity , we observed here with video-game play , although substantial , is somewhat smaller when compared with perceptual learning 4 ., However , it is not too surprising that direct training can produce greater improvements , as it usually involves a large number of practice trials ( for example , deep amblyopes might need more than 50 , 000 trials to reach the plateau levels 11 ) in which the task difficulty is very challenging , most of the time around the observers threshold limits ., In contrast to perceptual learning , video games provide a visually enriched and stimulating environment , demanding different fundamental visual skills ., Animal studies have highlighted the importance of environmental enrichment in promoting cortical plasticity 13 , 14 ., We postulate that the intense sensory-motor interactions while immersed in video-game play might push brain functions to the limit , enabling the visual system to learn , on the fly , to recalibrate and adjust , providing the basis for functional plasticity ., Treatment of adult amblyopia has recently received considerable attention ever since the introduction of perceptual learning techniques in the past few years 28–30 ., There have been numerous attempts to find an effective treatment for amblyopia ., These attempts include subcutaneous injection of strychnine 31 , flashing red and blue lights 32 , 33 , and rotating gratings 34 ., Other more recent studies have attempted to use electric stimulation 35 , direct transcranial magnetic stimulation 36 , and pharmacological approaches 37 to induce brain plasticity ., Some of these techniques seem promising , but the others lack repeatable clinical evidence ., Before a video-game-based approach is used to treat amblyopia clinically , there are still many questions to be addressed ( e . g . , dose-response , prognosis for different ages of onset , types and depths of amblyopia ) ., The current study serves as a “pilot” trial and , as such , has several design limitations: lack of randomization , small study size , and differences in numbers between arms ., The lack of randomization and differences in numbers between arms may have resulted in potentially imbalanced makeup of the study arms on baseline characteristics ., For example , the action game group was much more likely to be male and younger than the other groups ., In addition , the small number of participants ( four ) in the non-action game group makes it difficult to draw strong conclusions ., A much larger sample size is necessary for future studies to investigate which type , action or non-action , is more effective in treating amblyopia ., Specifically , a large-scale randomized double-blind clinical trial ( with equal numbers in each group ) is needed to eliminate differences between people , placebo effects , and measurement differences ., Despite these limitations , the present pilot study provides new insights into how video-game play sharpens visual functions in adult amblyopia and , most importantly , reveals that video-game play may provide important principles for improving treatment in amblyopia , and perhaps other clinical abnormalities ., The experimental procedures were approved by the University Committee for the Protection of Human Subjects , and the research was conducted according to the principles expressed in the Declaration of Helsinki ., Informed consent was obtained from each participant ., There was no known risk involved in the experimental procedures ., Altogether 20 adults with amblyopia participated in three video-game experiments ( age range: 15–61 y , mean age: 31 . 4±3 . 5 y ) ., They were recruited through advertisements in newspapers and through the Internet websites ., Thorough eye examination was carried out by an experienced optometrist ( first author , RWL ) ., Our participant inclusion criteria included: ( 1 ) age >15 years; ( 2 ) all forms of amblyopia , e . g . strabismic , anisometropic , refractive , deprivative , and meridional amblyopia; and ( 3 ) interocular visual acuity difference of at least 0 . 1 LogMAR ., Exclusion criteria included any ocular pathological conditions ( e . g . , macular abnormalities ) and nystagmus ., All of our participants had a difference in crowded visual acuity of two lines or more between the two eyes , and had normal vision in the sound eye ( ∼20/12–20/16 ) ., The maculae of all participants were assessed as normal , and they all had clear ocular media ( as assessed by direct ophthalmoscopy ) ., Their clinical data are summarized in Table, 1 . The study took place in our research laboratory at the University of California , School of Optometry in Berkeley , California , from December 2004 to December 2009 ., Participants were allocated into three intervention groups—two video-game treatment groups and one conventional occlusion therapy cross-over control group ( Figure 1a ) ., The first 10 enrolled patients participated in the action video game group , the subsequently enrolled three patients participated in the non-action videogame group , and then another seven patients were recruited in the cross-over intervention group of which participants were allowed to choose between the two types of video games ( MOH: n\u200a=\u200a4; SIM: n\u200a=\u200a1 , SC1 ) in phase, 2 . The two video games used were Medal of Honor Pacific Assault and SimCity Societies ( Electronic Arts , Inc . ) ., Since there has been no previous clinical evidence indicating that video games can modify vision in adult amblyopia in any way , in this pilot trial we decided to recruit participants for the video game treatment groups in the beginning , in order to evaluate the feasibility of this treatment approach ., It is important to note that the participant allocation was not based on the clinical characteristics of participants ., In the main experiments , participants were required to play the assigned video games in our research laboratory for 40 or 80 h ( 2 h/d ) using the amblyopic eye , with the fellow eye occluded with a black eye patch ., They were given full optical correction for the viewing distance ., A battery of vision function tests listed below was used to examine the effects of video-game experience on amblyopic vision ( Figure 1b ) ., All visual stimuli were displayed on a 21 in flat Sony F520 monitor screen at 1800×1440 resolution and 90 Hz refresh rate ., Not all participants completed every visual function testing ( visual acuity , n\u200a=\u200a20; positional acuity , n\u200a=\u200a16; visual counting , n\u200a=\u200a14; stereoacuity , n\u200a=\u200a5 ) ., Those participants in the control experiment ( OT group ) were given a log sheet to keep track of the patching hours and the visual tasks performed during patching . | Introduction, Results, Discussion, Materials and Methods | ClinicalTrials . gov NCT01223716 | Early abnormal visual experience disrupts neuronal circuitry in the brain and results in reduced vision , known as amblyopia or “lazy eye , ” the most frequent cause of permanent visual loss in childhood ., It is generally believed that adult amblyopia is irreversible beyond the sensitive period of brain development during childhood ., In this study , we examine whether playing video games , both action and non-action , has an effect on the vision of adults with amblyopia ., We assessed visual acuity ( visual resolution ) , positional acuity ( the ability to localize objects relative position ) , spatial attention ( the ability to direct visual attention to various locations in the visual field ) , and stereoacuity ( stereo-vision / 3-D depth perception ) in a small group of teenagers and adults ., We found that they tended to recover vision much faster than we would have expected from the results of conventional occlusion therapy in childhood amblyopia ., Additional experiments and modelling suggest that the improvements are a result of decreasing spatial distortion and increasing information processing efficiency in the amblyopic brain ., Thus , video games may include essential elements for active vision training to boost visual performance ., Most importantly , our findings suggest that video-game play may provide important principles for treating amblyopia , a suggestion that we are pursuing with larger scale clinical trials . | medicine, neuro-ophthalmology, cognitive neuroscience, pediatric ophthalmology, neurological disorders, neurology, ophthalmology, biology, sensory perception, neuroscience, learning and memory | A pilot study suggests that playing video games may enhance a range of spatial vision functions in adults with amblyopia. |
journal.pgen.1006449 | 2,016 | Genomic Characterization of Metformin Hepatic Response | Metformin is the first-line oral therapy for Type 2 Diabetes ( T2D ) 1 , and is also approved for use or used off-label in a variety of other diseases , such as polycystic ovary syndrome 2 , gestational diabetes 3 , pediatric obesity 4 and cancer 5 , 6 ., Side effects of metformin are mainly gastrointestinal in 20% to 30% of patients , and in very rare cases include lactic acidosis 7 ., However , the variability in response is substantial , with ≥30% of patients receiving metformin monotherapy classified as non-responders 8 ., The genomic characterization of metformin hepatic response would thus provide novel insights into the mechanisms of metformin action ., The molecular mechanisms of metformin action are not fully known 6 , 9 ., Metformin’s major tissue of action is the liver where it inhibits gluconeogenesis by activating the AMP-activated protein kinase ( AMPK ) pathway 10 , 11 ., Metformin-induced inhibition of the mitochondrial respiratory chain complex I leads to a reduction in ATP synthesis and to an increase in the cellular AMP:ATP ratio , which is thought to activate AMPK 12 ., Activation of AMPK is carried out by upstream kinases such as serine/threonine kinase 11 ( STK11/LKB1 ) , and ataxia telangiectasia mutated ( ATM ) that lead to AMPK phosphorylation in the presence of metformin 13 ., AMPK is also known to upregulate the nuclear receptor small heterodimer partner ( SHP ) upon metformin treatment 14 , which inhibits cAMP-response element-binding protein ( CREB ) -dependent hepatic gluconeogenic gene expression 12 , 15 ., Moreover , the phosphorylation of CREB binding protein ( CBP ) triggers the dissociation of transcription complexes that inhibit gluconeogenic genes 16 ., Metformin was also suggested to inhibit hepatic gluconeogenesis independent of the AMPK pathway , via a decrease in hepatic energy state through a process independent of the transcriptional repression of gluconeogenic genes 17 ., Moreover , it was proposed that metformin antagonizes the action of glucagon , thus reducing fasting glucose levels 18 ., Genetic variation can play an important role in metformin response , with a heritability of 34% based on genome-wide studies 19 ., Metformin is not metabolized and transporters are the major determinants of metformin pharmacokinetics ., Missense and promoter variants in transporter genes have been associated with metformin pharmacokinetics 20 , 21 ., Notably , genetic variants in OCT1 , the major determinant of metformin uptake in hepatocytes , have been associated with metformin action 22 , 23 ., Transcription factors that modulate the expression of metformin transporters were also associated with changes in metformin treatment outcome 24 ., A genome-wide association study ( GWAS ) found a noncoding single nucleotide polymorphism ( SNP ) rs11212617 nearby the ataxia telangiectasia mutated ( ATM ) gene to be associated with metformin treatment success 25 ., These results , though replicated in some smaller cohorts 26 , failed replication in a recent three-stage GWAS which identified an intronic SNP in the glucose transporter , SLCA2 , associated with better metformin response in multiple ethnically diverse cohorts 27 ., Combined , these studies have only been able to explain a small portion of the genetic variability associated with metformin response , suggesting a role for additional genetic determinants ., Here , we aimed to characterize metformin response pathways and regulatory elements in a systematic and genome-wide manner ., We performed RNA-seq and ChIP-seq for H3K27ac a known active mark 28 , and H3K27me3 a repressive mark 28 , on primary human hepatocytes from the same donor treated with the following three conditions:, 1 ) vehicle control ,, 2 ) metformin , and, 3 ) metformin and compound C ( an AMPK inhibitor 11 ) ., We identified AMPK-dependent and AMPK-independent gene clusters using RNA-seq ., We found thousands of peaks that were unique to cells treated with metformin using ChIP-seq ., Reporter assays in liver cells identified several promoters and enhancers that were modulated by metformin ., Moreover , enhancer assays of a metformin increased H3K27ac ChIP-seq peak that has SNPs in linkage disequilibrium ( LD ) with the metformin treatment response GWAS lead SNP rs11212617 25 , showed increased enhancer activity for the treatment response associate haplotype ., Expression quantitative trait locus ( eQTL ) liver analysis suggests that two SNPs within this enhancer are associated with increased ATM expression ., Using CRISPR activation ( CRISPRa ) , we found that in addition to ATM , EXPH5 and DDX10 could also be its target genes ., Further analysis of our top upregulated AMPK-dependent gene , activating transcription factor 3 ( ATF3 ) , using ChIP-seq and siRNA knockdown showed that it may have an important role in suppression of gluconeogenesis ., Our systematic studies highlight important metformin response associated genes and regulatory elements in the liver , providing novel sequence targets that could be associated with the vast variability in response to metformin , and identify novel T2D treatment candidates ., To comprehensively identify differentially expressed ( DE ) genes associated with metformin treatment , we carried out RNA-seq on primary human hepatocytes treated with:, 1 ) vehicle control ,, 2 ) 2 . 5 mM metformin for 8 hours , and, 3 ) 40 μM compound C , an inhibitor of AMPK 11 , along with 2 . 5 mM metformin for 8 hours ( Fig 1 ) ., Our RNA-seq analyses identified 1 , 906 DE genes between the vehicle control and both metformin treatment conditions using a p-value cutoff after correction for multiple testing of less than or equal to 0 . 05 ( S1 Table ) ., Amongst them , 1 , 255 were upregulated and 651 were downregulated ( Fig 2A ) ., Notably , we found novel transcription factors related to metformin response , such as the upregulated ATF3 and KLF6 and the downregulated AJUBA ( Fig 2B ) ., Ingenuity pathway analysis ( IPA ) found networks for upstream regulators enriched for DE genes , further implicating additional molecular pathways to metformin response ( S2 Table ) ., We also compared our RNA-seq data with previously reported microarray data from human hepatocytes treated with 1mM metformin for 8 hours 30 ., Despite the use of different techniques , conditions , statistical analyses and other variables that could confound these comparisons , we found that 25% of our DE genes overlap with microarray defined DE genes ( S2A Fig ) ., Moreover , we observed that several of the highly DE genes are similar in both datasets ( Fig 2C ) with fold changes showing a Spearman correlation of R2 = 0 . 52 ( S2B Fig ) ., We next generated AMPK-dependent and AMPK-independent clusters by comparing these three conditions ( Fig 2A ) ., Of note , as compound C is also thought to have off-target effects 31 , we only considered genes whose expression changed due to metformin response for this assay ., Clusters two ( n = 134 ) , three ( n = 575 ) , seven ( n = 83 ) and eight ( n = 168 ) contain genes whose expression increases with metformin treatment , but is greatly reduced with the combined treatment of compound C and metformin ., These genes were termed metformin increased , AMPK-dependent ., Gene ontology ( GO ) analysis found enrichment for transcription regulation and several additional terms for these clusters ( Fig 2A; S3 Table ) ., Cluster five ( n = 256 ) contains genes whose expression decreases with metformin treatment , but returns to untreated level with the combined treatment of compound C and metformin ., These genes were termed metformin decreased , AMPK-dependent and were enriched for ribonucleotide binding ( S3 Table ) ., In addition , IPA upstream regulator analysis of AMPK-dependent genes found many transcription factors not previously related to metformin and related to gluconeogenesis , and enrichment for the AMPK signaling canonical pathway ( Fig 3A; S3 Fig and S2 Table ) ., We also identified several AMPK-independent clusters ., Of note , Clusters one ( n = 194 ) , six ( n = 74 ) and ten ( n = 57 ) contain genes whose expression increases with metformin treatment that remains elevated with the combined treatment of compound C and metformin ( Fig 2A; S3 Table ) ., We termed these genes as metformin increased , AMPK-independent ., These genes were enriched for acute inflammatory response , chromatin organization and response to nutrient levels and other GO terms ( Fig 2A; S3 Table ) ., Clusters four ( n = 365 ) and nine ( n = 20 ) contain genes whose expression decreases with metformin treatment that remains low with the combined treatment of compound C and metformin ., These genes were termed metformin decreased , AMPK-independent and were enriched for metabolic processes , RNA processing and other GO terms ( S3 Table ) ., IPA upstream regulator analysis of AMPK-independent genes found transcription factors and ligand-dependent nuclear receptors not previously related to metformin , and enrichment for the acute phase response signaling canonical pathway ( Fig 3B; S3 Fig and S2 Table ) ., To identify metformin-responsive regulatory elements in a genome-wide manner , we performed ChIP-seq on primary human hepatocytes treated with the same three conditions in our RNA-seq ( vehicle control , metformin , compound C and metformin ) using antibodies for active ( H3K27ac ) and silenced ( H3K27me3 ) histone marks ( Fig 1A ) ., For H3K27ac , we annotated 19 , 232 peaks in non-treated cells compared to 21 , 120 upon metformin treatment and 32 , 311 peaks in cells treated with compound C and metformin ( Fig 1B; S4 Table ) , for a total of 34 , 910 distinct peaks across all conditions ., Of these peaks 12 , 847 where shared between the non-treated and metformin treated cells , while 14 , 211 were unique to the compound C and metformin treatment ( Fig 4A ) ., For the H3K27me3 repressive mark , we identified 20 , 844 peaks in non-treated cells compared to 45 , 909 upon metformin treatment , and 28 , 105 peaks with treatment of compound C and metformin ( Fig 1B; S4 Table ) ., The higher number of peaks identified in metformin treated cells ( 45 , 909 ) is likely due to technical differences , as the replicates for the other conditions showed larger variation between one another ., To gain an understanding of chromatin changes in response to metformin treatment , we used our ChIP-seq data to characterize the average changes in H3K27ac and H3K27me3 around transcription start sites ( TSSs ) of DE genes ., We observed that H3K27ac increased around the TSS of genes that were significantly upregulated by metformin , while H3K27me3 marks remained unchanged with no detectable signal in either condition ( Fig 4B ) ., However , among TSSs of significantly downregulated genes , H3K27ac coverage remained unchanged after treatment with metformin and H3K27me3 remained around zero in both conditions ., We also carried out an enrichment analysis between non-treated and metformin treated H3K27me3 peaks ., We did not identify any differentially enriched peaks , further suggesting that H3K27me3 marks are not responsive to metformin treatment , and differences in peak numbers may be due to ChIP efficiency ., Combined , these analyses suggest that H3K27ac undergoes more dynamic changes in response to metformin than H3K27me3 ( Fig 4C ) ., We next generated AMPK-dependent and AMPK-independent clusters of H3K27ac peak regions by comparing these three conditions ( Fig 4D ) ., Cluster one ( n = 1142 ) contained regions that showed increased H3K27ac marks upon treatment of metformin alone , which is then reduced with combined treatment of compound C . Similar to the RNA-seq analysis , these regions were termed metformin increased , AMPK-dependent ., Using the Genomic Regions Enrichment of Annotations Tool ( GREAT 32; Fig 4D; S5 Table ) we observed enrichment of peak regions nearby genes belonging to several GO categories , including negative regulation of protein phosphorylation and regulation of fatty acid metabolic process ., Additionally , peak regions belonging to this cluster where enriched for ATF3 binding motifs ( Fig 4D ) ., Cluster four ( n = 70 ) contains regions where H3K27ac histone modification decreases with metformin treatment , but returns to untreated levels with the combined treatment of compound C and metformin ., We also identified AMPK-independent H3K27ac peak region clusters ., As compound C is also thought to have off-target effects 31 , we only considered peaks whose expression changed due to metformin treatment for this assay ., Cluster two ( n = 87 ) contains peak regions whose H3K27ac enrichment increases with metformin treatment that remain elevated with the combined treatment of compound C and metformin ( S4 Table ) ., We termed these regions as metformin increased , AMPK-independent ., Cluster three ( n = 203 ) contain regions with H3K27ac histone modifications that decrease with metformin treatment and that remain low with the combined treatment of compound C and metformin ., These regions were termed metformin decreased , AMPK-independent and were associated with several metabolic pathways and xenobiotic stimulus ( S5 Table ) ., To functionally validate our ChIP-seq results , we tested putative promoter and enhancer sequences that had a H3K27ac peak upon metformin treatment ., We first analyzed the relative H3K27ac peak intensity between non-treated and metformin treated cells and identified 1 , 517 differentially enriched peaks ( S6 Table ) ., For promoter assays , we selected ten promoters of genes known to be involved in metformin response , several of which were enriched for H3K27ac upon metformin treatment ( S7 Table ) ., These sequences were cloned into a promoter assay vector ( pGL4 . 11b; Promega ) and tested for their promoter activity in Huh-7 liver cells treated with 2 . 5 mM metformin for 8 hours or vehicle control ., Among these promoters , seven exhibited promoter activity , and three were found to have differential activity upon metformin-treatment ( Fig 5A; S7 Table ) ., These three include the PFKFB2 promoter that was upregulated , and the PCK1 and SIRT1 promoters that were downregulated upon metformin treatment ., We performed quantitative PCR ( qPCR ) for these genes in Huh-7 cells and confirmed that PFKFB2 was increased by metformin ., However , PCK1 did not show a significant change and SIRT1 was actually increased in these assays ( Fig 5C ) ., Combined , these results suggest that potentially additional regulatory elements , such as enhancers , could control metformin response ., We next set out to test putative enhancer sequences for their activity using a similar luciferase reporter assay ., We selected fourteen putative enhancer sequences many of which had H3K27ac metformin enriched peaks and that reside near genes whose expression was induced by metformin or genes known to be associated with metformin response based on the literature ( S7 Table ) ., These sequences were cloned into an enhancer assay vector ( pGL4 . 23; Promega ) , which contains a minimal promoter followed by a luciferase reporter gene ., As a positive control , we used the ApoE liver enhancer 33 , whose activity should not be enhanced by metformin , and the pGL4 . 23 empty vector as a negative control ., All constructs were tested for their enhancer activity in Huh-7 cells treated with 2 . 5 mM metformin for 8 hour or vehicle control , similar to the promoter assays ., Out of the 14 assayed sequences , 8 showed significant enhancer activity and 4 were significantly induced upon metformin treatment ( Fig 5B; S7 Table ) ., These 4 metformin induced sequences reside near ATF3 , CRTC2 , NR0B2 and MYC genes ., In addition , a sequence near G6PC was significantly repressed upon metformin treatment ., We performed qPCR for these genes in Huh-7 metformin treated and untreated cells and found that ATF3 , NR0B2 and MYC were upregulated and G6PC was downregulated by metformin ( Fig 5C ) ., For the CRTC2 , we did not observe a significant upregulation , but did see a trend for metformin induction ( Fig 5C ) ., Our qPCR results overall agree with our enhancer assays , suggesting that these enhancers and genes could be regulated by metformin ., A noncoding SNP , rs11212617 , in the ATM locus was previously reported to be associated with metformin treatment success in a GWAS for glycemic response to metformin 25 ., To test for potential gene regulatory elements in this locus , we analyzed our ChIP-seq data for metformin enriched peaks in this locus ., We identified a metformin H3K27ac enriched peak in this region within an intron of the ATM gene that contains SNPs rs277070 and rs277072 that are in LD with rs11212617 R2>0 . 95 in the Caucasian ( CEU ) population ( Fig 6A ) ., We cloned both the unassociated and associated haplotype of this intronic peak into our enhancer assay vector and carried out similar luciferase reporter assays in Huh-7 cells as described above ., Both sequences showed significant enhancer activity , which was increased upon metformin response ., However , the treatment response associated haplotype showed significantly higher enhancer activity upon metformin response compared to the unassociated haplotype ( Fig 6B ) ., Our findings suggest that the haplotype that is in LD with the GWAS lead SNP rs11212617 increases enhancer activity and could lead to elevated expression of its target gene/s ., To identify potential target genes for this enhancer , we analyzed eQTL data and also took advantage of CRISPR/Cas9 activation technology ( CRISPRa; 34 ) to increase the activity of this enhancer and measure changes in expression levels of the nearby genes ., eQTL analysis of SNPs rs277070 and rs277072 ( see methods ) found significant associations with increased ATM mRNA expression for both SNPs ( S5 Fig; S8 Table ) ., For CRISPRa , we infected Huh-7 cells with a nuclease-deficient Cas9 ( dCas9 ) fused to the VP64 transcriptional activator along with three different single guided RNA ( sgRNA ) targeting the enhancer ., We generated mRNA and measured the expression levels of ATM in this region compared to cells with only dCas9-VP64 ., We found that ATM expression is significantly increased by about 2 fold by all three sgRNAs and when all 3 sgRNAs were infected together ( Fig 6C ) , with sgRNA1 and sgRNA3 providing the highest activation results ., We then measured the expression levels of additional genes in this locus Cullin 5 ( CUL5 ) , Acetyl-CoA Acetyltransferase 1 ( ACAT1 ) , Nuclear Protein , Ataxia-Telangiectasia Locus ( NPAT ) , Chromosome 11 Open Reading Frame 65 ( C11orf65 ) , DEAD ( Asp-Glu-Ala-Asp ) Box Polypeptide 10 ( DDX10 ) , finding that not only ATM , but also EXPH5 ( around 4 fold ) and DDX10 ( around 2 . 5 fold ) expression was significantly increased by both sgRNA1 and sgRNA3 ( Fig 6D ) ., Interestingly , EXPH5 and DDX10 ( but not ATM ) were upregulated by metformin in our RNA-seq data ( Fig 6E; S1 Table ) , which is consistent with the possible activation of these two genes by this enhancer ., Combined , these results suggest that this enhancer could be targeting either ATM , EXPH5 and/or DDX10 ., ATF3 was the top differentially expressed AMPK-dependent gene in our metformin treatment ( Fig 2B ) ., ATF3 is a member of the ATF/cyclic adenosine monophosphate ( cAMP ) responsive element-binding protein family of transcription factors , a stress response gene that both represses and activates genes35 , 36 ., ATF3 is known to be involved in metformin response in macrophages 37 , but its role in metformin response in the liver is not well characterized ., To better characterize the role of ATF3 in hepatic metformin response we carried out various assays including ATF3 ChIP-seq , siRNA and pathway analyses ., We first wanted to validate that ATF3 was indeed an AMPK-dependent gene , as observed in our RNA-seq results ., We thus treated Huh-7 cells with metformin , AICAR ( an AMPK activator ) , or compound C ( an AMPK inhibitor ) ., We found that ATF3 expression was induced by both metformin and AICAR treatment , but not with compound C ( Fig 7A ) , further confirming that ATF3 expression is dependent on AMPK , and that AMPK activation is required for ATF3 induction ., To obtain a genome-wide view of ATF3 function following metformin response , we carried out ATF3 ChIP-seq on primary human hepatocytes from the same individual used in our previous RNA-seq and ChIP-seq assays treated with 2 . 5mM metformin for 8 hours and compared them to vehicle control , similar to previous ChIP-seq experiments ., Since ATF3 is an AMPK-dependent gene , we did not carry out an additional compound C + metformin experiment ., We observed a massive recruitment of ATF3 binding across the genome following metformin treatment ., ATF3-bound DNA fragments clustered into 842 discrete peaks in non-treated cells compared to 3 , 535 upon metformin treatment ( Fig 1A , S4 Table ) , with 777 overlapping in both datasets ( Fig 7B ) ., Analysis of ATF3 metformin-treated unique peaks overlapping gene promoters found an enrichment of peaks occurring in DE genes as compared to all genes ( Fig 7C ) , including KLF6 , NR0B2 , DUSP1 , EIF2AK3 , IRS1 , NAMPT all known metformin-associated genes ., We also observed a peak overlapping the ATF3 promoter suggesting that it autoregulates itself ., Further analysis of the ATF3 peaks from metformin treated cells using both the Database for Annotation , Visualization and Integrated Discovery ( DAVID ) 38 and GREAT 32 found them to be enriched near genes involved in protein translation , noncoding RNA metabolic processes , RNA processing , cytoskeleton organization , transcription activation , ATP binding and many others ( Fig 7D and 7E; S9 Table ) ., In addition , we performed pathway analysis using IPA of genes found near ATF3-H3K27ac metformin peaks and found an enrichment for the top “EIF2 signaling” canonical pathway that is involved in protein translation ( S6 Fig; S10 Table ) ., We next set out to better characterize the functional role of ATF3 following metformin treatment ., Using IPA , we generated an ATF3 metformin response pathway centered on its interaction with metformin , AICAR and genes related to gluconeogenesis ( Fig 8A ) ., We then selected ATF3 regulated genes predicted by this pathway to be analyzed following ATF3 siRNA knockdown in Huh-7 cells ., We first validated that this siRNA can knockdown ATF3 , observing a 40% and 48% ATF3 mRNA reduction in non-treated and metformin treated cells respectively ( Fig 8B ) ., Next , we knocked down ATF3 in Huh-7 cells that were then treated with vehicle control or metformin , and performed qPCR on selected genes that were predicted by the ATF3 pathway ( Fig 8A ) ., In the basal condition , we observed significant downregulation following ATF3 knockdown for G6PC , JUN , MDM2 , PPP1R15A and SERPINE when compared to the siRNA control ( Fig 8B ) ., Upon metformin treatment , we observed a significant downregulation due to ATF3 knockdown for PPP1R15A and a significant upregulation of BIRC3 and PCK1 when compared to the negative siRNA control ( Fig 8B ) ., These results suggest that ATF3 activates PPP1R15A and represses BIRC3 and PCK1 upon metformin activation ( Fig 8A ) ., PPP1R15A codes for the protein phosphatase 1 , regulatory subunit 15A , which recruits the serine/threonine-protein phosphatase PP1 to dephosphorylate the translation initiation factor eIF-2A/EIF2S1 ., Interestingly , reducing the hepatic eIF2α signaling pathway in mice was shown to lead to reduced hepatic glucose production through reduced hepatic gluconeogenic gene expression 39 ., BIRC3 codes for the protein baculoviral IAP repeat containing 3 , and its expression was associated with the survival of insulin-secreting human liver cell line , which restored normoglycemia when transplanted into diabetic immunoincompetent mice 40 ., Furthermore , PCK1 encodes the phosphoenolpyruvate carboxykinase 1 ( PEPCK ) enzyme , which catalyzes the conversion of oxaloacetate to phosphoenolpyruvate and carbon dioxide and is considered a key pathway for hepatic gluconeogenesis 41 ., In combination , our findings suggest that ATF3 responds to metformin and could be involved in gluconeogenesis repression ., We have systematically characterized metformin hepatic response using RNA-seq and ChIP-seq on primary human hepatocytes treated with vehicle control , metformin , and compound C and metformin ., We identified AMPK-dependent and AMPK-independent clusters from 1 , 906 differentially expressed ( DE ) genes ., To identify metformin-associated regulatory elements , we carried out ChIP-seq for H3K27ac , a known active mark 28 , and H3K27me3 a repression mark 28 ., To functionally validate our results , we carried out functional reporter assays in Huh-7 cells with similar treatments finding several promoter and enhancer sequences to be modulated by metformin ., Moreover , enhancer assays of a metformin increased H3K27ac ChIP-seq peak in an ATM intron that contained SNPs in LD with the metformin treatment response GWAS lead SNP ( rs11212617 ) showed increased enhancer activity for the associated haplotype ., eQTL liver analysis suggests that SNPs within this enhancer are associated with increased ATM expression ., Using CRISPRa , we showed that this enhancer could be regulating either ATM , EXPH5 and/or DDX10 ., Finally , using ChIP-seq and siRNA , we characterized the metformin-associated function of ATF3 , the top upregulated AMPK-dependent gene , finding it to have an important role in gluconeogenesis repression ., Our RNA-seq analyses identified novel transcription factors associated with metformin response ., For example , krüppel-like factor 6 ( KLF6 ) was found among the upregulated AMPK-dependent genes ., KLF6 was implicated as a novel regulator of hepatic glucose and lipid metabolism in non-alcoholic fatty liver disease , characterized by dysregulated glucose homeostasis 42 ., Conversely , ajuba LIM protein ( AJUBA ) was found among the downregulated AMPK-independent genes ., It is thought to indirectly affect the activity of Protein Kinase Cζ 43 , which is required for LKB1 phosphorylation and , thus , AMPK activation 44 ., Notably , our IPA upstream regulator analysis for DE genes ( Fig 3 ) support that metformin affects the expression of many novel transcriptional regulators related to gluconeogenesis ., KLF6 isoforms were also found to be involved in tumor progression 45 , and the activity of AJUBA was found to promote cancer growth 46 ., Moreover , network analyses showed that CDKN1A , ESR1 , MAX , MYC , PPARGC1A , and SP1 play important roles in the antidiabetic and anticancer effects of metformin 47 ., Notably , these genes were also found to be DE in our RNA-seq ( S1 Table ) , and these findings support metformin as an emerging candidate for cancer therapy 5 , 6 ., RNA-seq analyses also identified novel protein coding genes associated with metformin response ., For example , dual specificity phosphatase 10 ( DUSP10 ) was found among the upregulated AMPK-dependent genes ( Fig 2B ) ., This gene negatively regulates members of the mitogen-activated protein kinases ( MAPK ) superfamily , some of them proposed as tumor suppressors 48 ., DUSP10 preferentially dephosphorylates p38 MAPK 49 ., Growth differentiation factor 15 ( GDF15 ) was the top upregulated AMPK-independent gene ( Fig 2B ) and its overexpression is known to result in improved insulin sensitivity 50 ., The transporter solute carrier family 19 , member 3 ( SLC19A3 ) was also an upregulated AMPK-dependent gene ( S1 Table ) , and is known to play a role in the intestinal absorption and tissue distribution of metformin 51 ., These findings further support that metformin induces protein coding genes which play a role in cancer and T2D ., Our promoter assays identified several functional promoters , but only a few were differentially regulated by metformin ., The promoter for PFKFB2 was induced by metformin , and showed increased expression by metformin both in our qPCR ( Fig 5C ) and RNA-seq ( S1 Table ) ., This gene is involved in both the synthesis and degradation of fructose-2 , 6-bisphosphate , a regulatory molecule that controls glycolysis ., However , we found discordant findings with the literature for other promoter assays ., For example , PCK1 codes for the PEPCK enzyme , a main control point for hepatic gluconeogenesis , which catalyzes the conversion of oxaloacetate to phosphoenolpyruvate and carbon dioxide 41 ., PCK1 was upregulated by metformin in our RNA-seq ( S1 Table ) , but the promoter for PCK1 was repressed by metformin , and PCK1 expression was decreased by metformin in Huh-7 cells ( Fig 5C ) ., Finally , SIRT1 codes for the sirtuin 1 protein , which leads to the suppression of gluconeogenic expression ., Metformin increases SIRT1 activity through activation of AMPK 52 ., SIRT1 showed increased expression by metformin both in our qPCR ( Fig 5C ) and RNA-seq ( S1 Table ) ., However , the promoter for SIRT1 was repressed by metformin ., These results could imply that although promoters can respond to metformin they may not be the only responders , suggesting a role for other metformin-responsive elements , such as enhancers ., Enhancers have been identified as potential determinants of drug response 53 , 54 ., Here , we used ChIP-seq for H3K27ac to identify metformin-responsive associated enhancers and found putative enhancer sequences near genes related to metformin action on gluconeogenesis ., Notably , an enhancer sequence near G6PC , which has a major role in gluconeogenesis 12 , was repressed by metformin ( Fig 5B ) ., Moreover , we found an enhancer that was induced by metformin near NR0B2 , which codes for the nuclear receptor SHP , a transcriptional repressor that inhibits CREB-dependent hepatic gluconeogenic expression via direct interaction and competition with coactivators 12 , 15 ., NR0B2 was also upregulated by metformin in our RNA-seq ( S1 Table ) and qPCR ( Fig 5C ) ., We also found a metformin inducible enhancer near CRTC2 , a transcription coactivator that is a key regulator of fasting glucose metabolism 55 ., It is worth noting that these sequences would not have been identified by conventional ChIP studies conducted in physiological conditions , nor would they be validated in functional assays without drug treatment , highlighting the need to carry out these drug-induced studies ., Together , our result suggests that ChIP-seq datasets are dependent on the environmental conditions in which they were performed , and that there are likely many enhancers which are only observed following a specific stimulus ., We also followed up on an ATM intronic enhancer that encompassed SNPs in LD with a GWAS SNP that was associated with metformin treatment success 25 ., Enhancer assays showed increased enhancer activity upon metformin response for the associated haplotype ( Fig 6B ) , suggesting that it could increase enhancer activity and elevate the expression of its target gene/s ., eQTL analysis showed that the SNPs within this enhancer are associated with increased ATM expression ., Next , we used CRISPRa to further identify the target gene/s ., Although ATM expression was increased by three sgRNA targeting this enhancer region , this gene may not be the only gene regulated by this enhancer region ., EXPH5 and DDX10 expression were also significantly increased ., Interestingly , EXPH5 and DDX10 , but not ATM , were upregulated by metformin in our RNA-seq , which are consistent with the possible metformin induction of these two genes by this enhancer ., The SNP rs11212617 falls within a large block of LD including CUL5 , ACAT1 , NPAT , ATM , C11orf65 , KDELC2 , EXPH5 25 ., Interestingly , we noted from GRASP v2 . 0 database 56 that SNPs near DDX10 are associated with fasting insulin and HOMA-IR ( p<3x10-5 ) 57 , and SNPs near EXPH5 are associated with fasting glucose ( p<3x10-3 ) 58 ., However , to our knowledge , no previous study has related EXPH5 or DDX10 to metformin response ., ATF3 is a transcription factor which can act as either a tra | Introduction, Results, Discussion, Methods | Metformin is used as a first-line therapy for type 2 diabetes ( T2D ) and prescribed for numerous other diseases ., However , its mechanism of action in the liver has yet to be characterized in a systematic manner ., To comprehensively identify genes and regulatory elements associated with metformin treatment , we carried out RNA-seq and ChIP-seq ( H3K27ac , H3K27me3 ) on primary human hepatocytes from the same donor treated with vehicle control , metformin or metformin and compound C , an AMP-activated protein kinase ( AMPK ) inhibitor ( allowing to identify AMPK-independent pathways ) ., We identified thousands of metformin responsive AMPK-dependent and AMPK-independent differentially expressed genes and regulatory elements ., We functionally validated several elements for metformin-induced promoter and enhancer activity ., These include an enhancer in an ataxia telangiectasia mutated ( ATM ) intron that has SNPs in linkage disequilibrium with a metformin treatment response GWAS lead SNP ( rs11212617 ) that showed increased enhancer activity for the associated haplotype ., Expression quantitative trait locus ( eQTL ) liver analysis and CRISPR activation suggest that this enhancer could be regulating ATM , which has a known role in AMPK activation , and potentially also EXPH5 and DDX10 , its neighboring genes ., Using ChIP-seq and siRNA knockdown , we further show that activating transcription factor 3 ( ATF3 ) , our top metformin upregulated AMPK-dependent gene , could have an important role in gluconeogenesis repression ., Our findings provide a genome-wide representation of metformin hepatic response , highlight important sequences that could be associated with interindividual variability in glycemic response to metformin and identify novel T2D treatment candidates . | Metformin is among the most widely prescribed drugs ., It is used as a first line therapy for type 2 diabetes ( T2D ) , and for additional diseases including cancer ., The variability in response to metformin is substantial and can be caused by genetic factors ., However , the molecular mechanisms of metformin action are not fully known ., Here , we used various genomic assays to analyze human liver cells treated with or without metformin and identified in a genome-wide manner thousands of differentially expressed genes and gene regulatory elements affected by metformin ., Follow up functional assays identified several novel genes and regulatory elements to be associated with metformin response ., These include ATF3 , a gene that showed gluconeogenesis repression upon metformin response and a potential regulatory element of the ATM gene that is associated with metformin treatment differences through genome-wide association studies ., Combined , this work identifies several novel genes and gene regulatory elements that can be activated due to metformin treatment and thus provides candidate sequences in the human genome where nucleotide variation can lead to differences in metformin response ., It also enables the identification and prioritization of novel candidates for T2D treatment . | genome-wide association studies, medicine and health sciences, liver, evolutionary biology, gene regulation, regulatory proteins, population genetics, dna-binding proteins, regulator genes, genome analysis, transcription factors, gene types, population biology, small interfering rnas, genomics, animal cells, proteins, gene expression, hepatocytes, biochemistry, rna, haplotypes, cell biology, nucleic acids, anatomy, genetics, biology and life sciences, cellular types, computational biology, non-coding rna, human genetics | null |
journal.pgen.1000074 | 2,008 | A Genome-Wide Association Study Identifies Novel Alleles Associated with Hair Color and Skin Pigmentation | There is substantial variation in human pigmentation within and across populations ., Ultraviolet radiation ( UV ) exposure is the most important environmental factor influencing evolutionary selection pressure on pigmentation ., In addition to UV-induced DNA damage , UVA can break down folic acid , and the major source of circulating vitamin D is synthesized in UVB-exposed skin ., Because both nutrients are essential for human reproduction , it has been proposed that human pigmentation is selected , at least in part , to optimize levels of these two UV-related nutrients 1 ., UV light is also the major environmental risk factor for skin cancer in humans ., Red and blonde hair color , light skin pigmentation , and blue eye color are major host susceptibility factors for skin cancer 2 ., Human pigmentation is a polygenic quantitative trait with high heritability 3–5 ., A handful of genes underlying rare , extreme pigmentation phenotypes have been discovered 6 , although until recently , only six genes were known to contain common genetic variants associated with human pigmentation in the normal range ( MC1R , TYR , OCA2 , SLC24A5 , MATP , and ASIP ) ., The proteins that these genes encode contribute to the control of melanin production and the maturation of melanosomes in melanogenesis , which determines human pigmentation ., With new technologies that enable genotyping of hundreds of thousands of single nucleotide polymorphisms ( SNPs ) , together with new insights into the structure of variation in the human genome 7 , it is now possible to scan the genome in an agnostic manner in search of common genetic variants associated with human pigmentation ., To identify common genetic variants associated with variation in natural diversity of human pigmentation , we performed a genome-wide association study ( GWAS ) of natural hair color in 2 , 287 U . S . women of European ancestry using data on 528 , 173 SNPs genotyped as part of the Cancer Genetic Markers of Susceptibility breast cancer GWAS 8 ., Promising SNPs were examined in four additional studies with data on hair color and other pigmentation phenotypes: 870 U . S . women controls free of diagnosed skin cancer from a skin-cancer case-control study; 3 , 750 U . S . women from a diabetes case-control study; 2 , 405 U . S . men from a diabetes case-control study; and 1 , 440 parents of twins from an ongoing Australian family-based study of genetic and environmental factors contributing to the development of pigmented nevi 9 ., The frequencies of pigmentary phenotypes collected in the 5 component studies are presented in Table, 1 . The samples were broadly similar ., We compared the distribution of observed p-values from each of the 528 , 173 SNPs in the GWAS with those expected under the global null hypothesis that none of the tested SNPs is associated with natural hair color ( Figure 1 ) ., The distribution of the observed p-values for the crude analyses that restricted analysis to women of self-reported European ancestry but did not further adjust for potential population stratification shows evidence for systematic bias: the genomic control inflation factor for the crude analyses ( the ratio of the median observed test statistic to the theoretical median ) is λGC\u200a=\u200a1 . 24 ., This systematic bias is most likely due to confounding by latent population stratification ., Hair color varies along a light-dark gradient from northern to southern Europe , so it will be associated with any SNP marker whose minor allele frequency also varies along a North-South gradient , even if that marker is not in linkage disequilibrium ( LD ) with a causal hair-color locus 10 ., Adjusting for the top four principal components of genetic variance 11 eliminated most of the apparent residual confounding due to population stratification ( λGC\u200a=\u200a1 . 02 for the adjusted analyses ) ; further control for up to 50 principal components did not alter the λGC ., All of the association results from the initial GWAS reported below are from analyses that adjusted for the top four principal components of genetic variation ., The GWAS identified several genomic locations as potentially associated with hair color ( Figure 2 ) ., Of 528 , 173 SNPs tested , the 38 SNPs with the most extreme p-values associated with hair color are listed in Table, 2 . We selected 31 of these 38 SNPs for further study in an independent sample ., The remaining seven SNPs were in strong LD ( r2>0 . 8 ) with one of these 31 SNPs ( Table 2 ) ., The sample consisted of 870 controls of European ancestry from a nested case-control study of skin cancer within the Nurses Health Study ( NHS ) ., Thirty of the 31 attempted SNPs were genotyped successfully ., Twenty-two of these 30 SNPs showed very strong evidence for association with natural hair color ( p<9 . 5×10−8\u200a=\u200a0 . 05/528 , 173 ) in a pooled analysis of the initial GWAS and the validation sample ( Table 3 ) ., Of the remaining eight SNPs , three showed very strong evidence for association with hair color either after excluding women with red hair or when comparing women with red hair to those without ( Table 3 ) ., The associations between these 30 SNPs and with tanning ability and skin color are presented in Table S1 and Table S2 ., The SNP rs12203592 in intron 4 of the IRF4 gene was strongly associated with hair color in the initial GWAS and validation study ( black to red , pooled p value for trend\u200a=\u200a8 . 5×10−28; black to blonde , pooled p value for trend\u200a=\u200a7 . 1×10−49 ) ., The percentage of residual variation in hair color from black to blonde explained by this SNP after controlling for the top four principal components of genetic variation was 7 . 0% ., This SNP is within 69 . 7 kb of two SNPs ( rs4959270 and rs1540771 ) that were identified by a recent GWAS of natural hair color in women of European ancestry resident in Iceland 12 ., However , neither of these variants , which lie between EXOC2 ( SEC5L1 ) and IRF4 , was as strongly associated with natural hair color in our initial GWAS as the IRF4 SNP rs12203592 ( Figure 3 ) ., In our GWAS , the p values for association between hair color ( black to blonde ) and rs4959270 and rs1540771 were 2 . 9×10−4 and 0 . 007 , respectively , and those for tanning ability were 0 . 002 and 0 . 001 , respectively ., In fact , the p-value for association between rs12203592 and natural hair color was more than 13 orders of magnitude smaller than the p-value for any other SNP on chromosome 6 ., This should not be taken as evidence that the loci that influence hair color in Iceland are different from those for the rest of Europe; rather , the previous GWAS may have failed to identify rs12203592 because this SNP is not on the Illumina HumanHap300 array used in that study , while it is on the Illumina HumanHap550 array used here ., We genotyped rs12203592 in an additional 6 , 155 individuals of predominantly European ancestry from the United States , including 3 , 750 women from the NHS and 2 , 405 men from the Health Professionals Follow-up Study ( HPFS ) , and in an additional 1440 individuals of European ancestry from Australia ( Queensland Institute of Medical Research ( QIMR ) ) ., The association with hair color showed strong reproducibility; the independent p-values for association with hair color ( black to blonde ) in these three follow-up studies was 3 . 2×10−40 , 2 . 8×10−35 , and 4 . 6×10−23 respectively , and the pooled p-value across all five studies was 1 . 5×10−137 ( Table 4 ) ., There was no statistical evidence for heterogeneity in the magnitude or direction of the hair color-minor allele correlation across the studies ., The rs12203592 SNP was also highly associated with skin color ( 6 . 2×10−14 ) , eye color ( 6 . 1×10−13 ) , and tanning ability ( 3 . 9×10−89 ) in subsets of these individuals for which this information was available ( Table 4 ) ., This variant allele was associated with lighter skin color , less tanning ability , and blue/light eye color ( Table 5 ) ., On the same chromosome , 145 . 8 kb centromeric from the IRF4 rs12203592 , the SNP rs6918152 in the EXOC2 gene was associated with hair color ( black to blond ) in the initial GWAS , the NHS skin cancer controls , and the Australian samples ( Tables 3 and 6 ) ., These two SNPs are in very weak LD ( r2\u200a=\u200a0 . 04 ) ., Genotypes for the SNP rs1540771 previously reported by Sulem et al . 12 were available in the initial scan and the Australian samples ., In a mutually adjusted multivariable regression of rs12203592 , rs6918152 , rs1540771 , the strength of the association between the first SNP with hair color was attenuated but remained significant ( p\u200a=\u200a2 . 7×10−17 for IRF4 rs12203592 in the Australian samples ) ( Table 6 ) ., The previously reported SNP rs1540771 was not significant ( p>0 . 05 ) after adjustment for the other SNPs , and association between rs6918152 and hair color was much weaker in the GWAS and no longer significant in the Australian samples after adjustment ., While the IRF4 SNP rs12203592 was also associated with skin color , eye color and tanning ability , the EXOC2 SNP rs6918152 was not associated with these phenotypes ., These results suggest that the IRF4 SNP rs12203592 is most likely to be in strong LD with the causal variant in this region ., The rs12896399 SNP 15 . 5 kb upstream of the SLC24A4 gene was highly associated with light hair color , and relatively weakly associated with less tanning ability in the pooled analysis of four studies ( p\u200a=\u200a6 . 0×10−62 for hair color , and p\u200a=\u200a0 . 01 for tanning ability ) ., The percentage of residual variation in hair color from black to blonde explained by this SNP after controlling for the top four principal components of genetic variation was 2 . 6% ., This variant was also associated with blue/light eye color ( p\u200a=\u200a2 . 9×10−6 in the HPFS set ) ., The SLC24A4 gene belongs to a family of potassium-dependent sodium/calcium exchangers ., At least two other members of this family are associated with skin pigmentation ., The SLC24A5 gene was recently shown to be involved in skin pigmentation in both zebrafish and humans 13 ., Another member of this family , MATP ( SLC45A2 ) , is a pigmentation gene transcriptionally regulated by MITF 14 , 15 ., We identified the SNP rs28777 in the MATP gene from the GWAS , and the association with hair color was replicated in the controls of the skin cancer study ( pooled P value\u200a=\u200a8 . 9×10−14 ) ., This SNP was also associated with skin color ( pooled P value\u200a=\u200a9 . 5×10−4 ) and tanning ability ( pooled P value\u200a=\u200a2 . 2×10−10 ) ., Three SNPs in the MATP gene have been associated with human pigmentation: rs16891982 ( Phe374Leu ) , rs26722 ( Glu272Lys ) , and rs13289 C/G ( -1721 in the promoter region ) 16 , 17 ., We genotyped these three SNPs in the controls of the skin cancer study ., None of the three previously reported SNPs were in LD with rs28777 ( r2≤0 . 01 ) , which is an intronic SNP ., A multivariable analysis mutually adjusting for rs28777 , rs16891982 , rs26722 , and rs13289 simultaneously showed that only rs16891982 remained significant in the model ( P\u200a=\u200a0 . 036 for hair color ( black to blonde ) , P\u200a=\u200a0 . 016 for tanning ability , and P\u200a=\u200a0 . 0009 for skin color ) and other SNPs became non-significant ( p>0 . 05 ) ., These data suggested that rs16891982 is most likely to be the causal variant or in strong LD with the causal variant in the MATP gene ., Eleven SNPs spanning 1 Mb on chromosome 15 were strongly associated with hair color in the initial GWAS ., These SNPs were located in the OCA2 5′ regulatory region and the HERC2 gene region and included the 3 SNPs reported previously with eye color: rs7495174 , p\u200a=\u200a7 . 1×10−7; rs6497268 , p\u200a=\u200a8 . 5×10−15; and rs11855019 , p\u200a=\u200a2 . 1×10−12 ) 18 ., In an analysis mutually adjusting for all 11 SNPs simultaneously , only the HERC2 SNP rs12913832 ( not on the HumanHap 300 version used in Sulem et al . 12 ) remained significantly associated with hair color ( p\u200a=\u200a2 . 73×10−32 ) and tanning ability ( p\u200a=\u200a3 . 03×10−9 ) ., The associations between all other SNPs and hair color became non-significant ( p>0 . 05 ) ., This suggested that the SNP rs12913832 was most likely to be in strong linkage disequilibrium with the causal variant in this region ., The percentage of residual variation in hair color from black to blonde explained by this SNP after controlling for the top four principal components of genetic variation was 10 . 7% ., We observed 12 SNPs on chromosome 16 associated with hair color in the GWAS , spanning >756 kb ., The MC1R gene , well established to be associated with red hair color , is located within this region ., We had previously genotyped 7 common MC1R variants among the NHS skin cancer controls 19 ., The analysis mutually adjusting for all 19 SNPs in the controls of the skin cancer study indicates that the signals that we detected in this region were mainly due to the three MC1R red hair color alleles ( Arg151Cys , Arg160Trp , and Asp294His ) ( Table S3 ) ., The pairwise LD among these 19 SNPs was very low ( the pattern of LD across these 19 SNPs is shown in Figure S1 ) ., It has been a longstanding hypothesis that human pigmentation is tightly regulated by genetic variation ., However , very few genes have been identified that contain common genetic variants associated with human pigmentation ., We conducted a genome-wide association study of hair color and identified several new variants associated with variation in hair color , skin color , eye color , and tanning ability among individuals of European ancestry ., Among the loci identified from our GWAS , the IRF4 and SLC24A4 loci had not been linked to human pigmentation before we began our study ., Recently , Sulem et al . 12 reported a pigmentation GWAS using 316 , 515 SNPs in the Icelandic population ., The associations of hair color in our GWAS with the 60 SNPs reported by Sulem et al . are listed in Table S4 ., These authors identified two SNPs ( rs4959270 and rs1540771 ) between the EXOC2 and IRF4 genes in relation to freckles , hair color , and skin sensitivity to sun 12 ., We identified a SNP in intron 4 of the IRF4 gene , not genotyped by Sulem et al . 12 , with a much stronger association than they observed with hair color and tanning ability ., The IRF4 gene product is a member of the interferon regulatory factor family of transcription factors 20–23 , which are involved in the regulation of gene expression in response to interferon and other cytokines ., The IRF4 gene encodes a B-cell proliferation/differentiation protein , which has been proposed as a sensitive and specific marker for conventional primary and metastatic melanomas and benign melanocytic nevi 24 ., The SNP rs12896399 upstream of the SLC24A4 gene showed strong association with hair color in our study and that of Sulem et al . 12 ., In addition , we identified three chromosomal regions adjacent to the previously known pigmentation genes: MC1R , OCA2 , and MATP ., The MC1R gene encodes a 317-amino acid 7-pass-transmembrane G protein coupled receptor and has been shown as the rate-limiting step in the activation of the cAMP pathway in terms of melanin production ., Although the LD between the MC1R variants and surrounding highly significant SNPs was relatively low , the multivariable models mutually adjusting for all surrounding SNPs suggests that the signals that we identified on chromosome 16 were explained by the functional variants in the MC1R gene ., A previous report showed that three SNPs in intron 1 of the OCA2 gene were associated with eye , skin , and hair color 18 ., We identified a SNP ( rs12913832 ) in the upstream HERC2 gene with a much stronger association with hair color and tanning ability ., Because the HERC2 gene has not been linked to human pigmentation to our knowledge , the SNP may be involved in the regulation of the expression or the function of the OCA2 gene ., Similarly , Sulem et al . identified rs1667394 ( ∼165 kb upstream from rs12913832 , r2\u200a=\u200a0 . 58 ) as the strongest hit in this region in their study 12 ., However , we found a stronger association of rs12913832 with hair color than that of rs1667394 in our study ., Similar to our findings , Sulem et al . reported associations in the regions encompassing MC1R , OCA2 , and SLC24A4 ., In addition , they reported loci in two established pigmentation genes , TYR and KITLG , which were not among the 38 SNPs with the strongest associations with hair color ( black to red ) that we sought to genotype in additional samples ., The p-value for the association between KITLG rs12821256 and hair color ( black to red ) in the initial GWAS was 0 . 0002; the p-value for the association with hair color excluding women with red hair ( black to blonde ) was 1 . 28×10−8 ., The p-values for association between TYR rs1393350 and hair color coded as black to red or black to blonde were 0 . 05 and 0 . 02 , respectively ., We additionally identified another previously reported pigmentation gene from our GWAS , the MATP gene that was not reported by Sulem et al . 12 ., In our analysis of testing the trend across hair color from black to blonde , there were no other loci reaching genome-wide significance level ., Two of the four regions we found to be associated with variation in hair color among Europeans without red hair ( MATP and HERC2/OCA2 ) show strong evidence of recent positive selection , based on a comparison of allele frequencies across samples from three continental populations ( Africa , Asia , and Europe ) 25 ., Both of the markers we identified in the two remaining regions showed significant differences in allele frequency across the HapMap CEU , CHB , JPT and YRI panels: the IRF4 SNP rs12203592 was monomorphic in CHB , JPT and YRI panels ( the minor allele in Europeans was absent from these samples ) ; the minor allele among Europeans for SLC24A4 SNP rs12896399 ( G ) was the major allele for the CHB and YRI panels , with the G allele frequency in the YRI sample being above 99% ., Moreover , all of the markers with strongest association with hair color in these four regions were significantly associated with one or more of the top four principal components of genetic variation ( Table S5 ) , suggesting that allele frequencies for these markers also vary among European populations ., Because we adjusted for latent population structure using these four principal components—and there are multiple lines of evidence suggesting these regions influence hair color among Europeans—we believe it unlikely that the strong associations we see between these markers and hair color are solely due to population stratification bias ., Rather it is likely that differences in the distribution of hair color across Europe are due in part to differences in allele frequencies at these loci and other as-yet-unknown loci ., Taken together , these four regions explain approximately 21 . 9% of the residual variation in hair color ( black-blond ) after adjusting for the top four principal components of genetic variation ., ( Conversely , after adjusting for these four regions , the top four principal components of genetic variation explain 2 . 6% of the residual variation in hair color . ), In our study of men and women of European ancestry we focused on the most statistically significant associations from our GWAS among women , identifying the IRF4 variant as reproducibly associated with human pigmentation ., Further work is needed to identify the causal variant at this locus ., Because a subset of true associations would be weakly associated with outcome in any given GWAS , large-scale replication is necessary for confirmation , and some true associations may be missed if they are not carried forward into replication studies ., In this regard , the precomputed rankings and P values for all the SNPs included in the GWAS conducted in the NHS are freely available ( http://www . channing . harvard . edu/nhs/publications/index . shtml ) for others to use in subsequent studies ., The NHS was established in 1976 , when 121 , 700 female U . S . registered nurses between the ages of 30 and 55 , residing in 11 larger U . S . states , completed and returned the initial self-administered questionnaire on their medical histories and baseline health-related exposures , forming the basis for the NHS cohort ., Biennial questionnaires with collection of exposure information on risk factors and ( every 4 years since 1980 ) nutritional assessments have been collected prospectively ., Along with exposures every 2 years , outcome data with appropriate follow-up of reported disease events , including melanoma and non-melanoma skin cancers , are collected ., Overall , follow-up has been very high; after more than 20 years approximately 90% of participants continue to complete questionnaires ., From May 1989 through September 1990 , we collected blood samples from 32 , 826 participants in the NHS cohort ., Subsequent follow-up has been greater than 99% for this subcohort ., The information on natural hair color at age of 20 and childhood and adolescence tanning ability were collected in the 1982 questionnaire ., We initially performed genotyping in a nested case-control study of postmenopausal invasive breast cancer within the Nurses Health Study ( NHS ) cohort 26 using the Illumina HumanHap550 array , as part of the National Cancer Institutes Cancer Genetic Markers of Susceptibility ( CGEMS ) Project 8 ., We performed our initial genome-wide analysis on 528 , 173 SNPs in 2 , 287 women 8 ., All cases and controls were self-described as being of European ancestry ., Four samples were excluded because of evidence of intercontinental admixture ., Controlling for breast cancer case-control status made no material difference to the GWAS results ., Information on natural hair color at age 20 was collected in the NHS main questionnaire and grouped into five categories ( black , dark brown , light brown , blonde , and red ) ., Detailed methods related to the initial GWAS were published previously 8 , including genotyping and quality control , initial assessment of sample completion rates , assessment of SNP call rates , concordance rate , deviation from Hardy–Weinberg proportions in control DNA , and final sample selection and exclusion for association analysis ., The promising SNPs from the initial GWAS were further genotyped among 870 controls in the skin cancer nested case-control study within the NHS ., The distribution of risk factors for skin cancer in the subcohort of those who donated blood samples was very similar to that in the overall cohort 2 ., A common control series was randomly selected from participants who gave a blood sample and were free of diagnosed skin cancer up to and including the questionnaire cycle in which the corresponding case was diagnosed ., In 1986 , 51 , 529 men from all 50 U . S . States in health professions ( dentists , pharmacists , optometrists , osteopath physicians , podiatrists , and veterinarians ) aged 40–75 answered a detailed mailed questionnaire , forming the basis of the study ., Between 1993 and 1994 , 18 , 159 study participants provided blood samples by overnight courier ., The information on natural hair color and eye color was collected in the 1988 questionnaire , and the information on tanning ability was asked in the 1992 questionnaire ., Two additional studies were used to genotype novel pigmentation loci: 3 , 750 samples from the nested case-control study of diabetes in the NHS and 2 , 405 samples from the nested case-control study of diabetes in the HPFS 27 ., All samples that we used were cases and controls from these two studies ., Cases were incident cases of diabetes after blood collection , and controls were matched on age ., Controlling for case-control status made no material difference to the results ., There was no sample overlap among the initial GWAS , the skin cancer case-control study , and the two diabetes case-control studies ., The study protocol was approved by the Institutional Review Board of Brigham and Womens Hospital and Harvard School of Public Health ., Informed consent was obtained from all patients ., The Australian sample comprised 1 , 442 parents of twins taking part in a long-running study of melanoma risk factors 9 , 18 , 28 , 29 ., Participants rated their own hair color ( at age 20 years ) on a five-point classification ( blonde , light brown , dark brown , black , red ) , eye color ( blue/grey , green/hazel , brown ) , and skin color ( light , medium , or dark ) ., For the primary analysis of hair color we regressed an ordinal coding for hair color ( 1\u200a=\u200ablack; 2\u200a=\u200adark brown; 3\u200a=\u200alight brown; 4\u200a=\u200ablonde; and 5\u200a=\u200ared ) on an ordinal coding for genotype ( 0 , 1 or 2 copies of SNP minor allele ) separately for each SNP that passed quality control filters 8 ., Crude analyses that did not adjust for any other variables showed evidence of systematic bias ( see results ) ; as this bias was greatly reduced by adjusting for the four largest principal components of genetic variation , all subsequent association analyses in the initial GWAS included these four components in the regression model ., These principal components were calculated for all individuals on the basis of ca ., 10 , 000 unlinked markers using the EIGENSTRAT software 8 , 11 ., The top four eigenvectors were chosen on the basis of significant ( p<0 . 05 ) Tracy-Wisdom tests 30 ., Adjusting for up to the top 50 principal components did not further reduce the genomic control inflation factor λGC ., We chose markers for genotyping in subsequent validation studies based on the p-values for association from the primary analysis ., Partial correlation coefficients ( i . e . , adjusted r2 or the percent of residual variance explained by the SNP marker ) were calculated from the linear regressions adjusted for the top four principal components of genetic variation ., There is some evidence that determinants of hair color may act along two phenotypic axes: red hair color versus non-red color and light-dark variation among those without red hair ., For example , alleles at the MC1R locus primarily determine presence or absence of red hair 31 ., Hence , we conducted further analyses among individuals without red hair and comparing those with red hair to those without to evaluate whether discovered loci act on one or both phenotypic axes ., We regressed the ordinal coding for hair color on minor allele counts excluding the individuals with red hair and used logistic regression to test the association between the ordinal genotype coding and a binary red-hair phenotype ( red vs . non-red hair color ) ., The regression parameter beta refers to the mean change in hair color scoring ( or change in log odds of red hair for red hair analyses ) per copy of the SNP minor allele ., We also used linear regression to test association between minor allele counts and self-reported tanning in response to sunlight ( 1\u200a=\u200adeep tan , 2\u200a=\u200aaverage tan , 3\u200a=\u200alight tan , 4\u200a=\u200ano tan ) , eye color ( 1\u200a=\u200abrown/dark , 2\u200a=\u200ahazel/green/medium , and 3\u200a=\u200ablue/light ) , and skin color ( 1\u200a=\u200ablack , 2\u200a=\u200amedium , and 3\u200a=\u200afair ) ., Pooled analyses of multiple studies were conducted by merging data sets and including separate baseline parameters for each study ., The TaqMan/BioTrove assays on the 31 SNPs in the skin cancer controls were performed at the Dana Farber/Harvard Cancer Center Polymorphism Detection Core ( primers and probe sequences are available on request ) ., Two loci ( IRF4 rs12203592 and SLC24A4 rs12896399 ) were further genotyped in diabetes samples in the NHS and HPFS studies using the Taqman assay ., Laboratory personnel were blinded to the case-control status , and 10% blinded quality control samples were inserted to validate genotyping procedures; concordance for the blinded samples was 100% ., Primers , probes , and conditions for genotyping assays are available upon request ., For the Australian study , genotyping was performed as a single multiplex reaction on the Sequenom high-throughput genotyping platform on IRF4 SNP rs12203592 and EXOC2 rs6918152 and the best pigmentation-associated SNPs in the region of 6p25 ( rs1540771 ) reported by Sulem et al . 12 . | Introduction, Results, Discussion, Materials and Methods | We conducted a multi-stage genome-wide association study of natural hair color in more than 10 , 000 men and women of European ancestry from the United States and Australia ., An initial analysis of 528 , 173 single nucleotide polymorphisms ( SNPs ) genotyped on 2 , 287 women identified IRF4 and SLC24A4 as loci highly associated with hair color , along with three other regions encompassing known pigmentation genes ., We confirmed these associations in 7 , 028 individuals from three additional studies ., Across these four studies , SLC24A4 rs12896399 and IRF4 rs12203592 showed strong associations with hair color , with p\u200a=\u200a6 . 0×10−62 and p\u200a=\u200a7 . 46×10−127 , respectively ., The IRF4 SNP was also associated with skin color ( p\u200a=\u200a6 . 2×10−14 ) , eye color ( p\u200a=\u200a6 . 1×10−13 ) , and skin tanning response to sunlight ( p\u200a=\u200a3 . 9×10−89 ) ., A multivariable analysis pooling data from the initial GWAS and an additional 1 , 440 individuals suggested that the association between rs12203592 and hair color was independent of rs1540771 , a SNP between the IRF4 and EXOC2 genes previously found to be associated with hair color ., After adjustment for rs12203592 , the association between rs1540771 and hair color was not significant ( p\u200a=\u200a0 . 52 ) ., One variant in the MATP gene was associated with hair color ., A variant in the HERC2 gene upstream of the OCA2 gene showed the strongest and independent association with hair color compared with other SNPs in this region , including three previously reported SNPs ., The signals detected in a region around the MC1R gene were explained by MC1R red hair color alleles ., Our results suggest that the IRF4 and SLC24A4 loci are associated with human hair color and skin pigmentation . | It has been a longstanding hypothesis that human pigmentation is tightly regulated by genetic variation ., However , very few genes have been identified that contain common genetic variants associated with human pigmentation ., We scanned the genome for genetic variants associated with natural hair color and other pigmentary characteristics in a multi-stage study of more than 10 , 000 men and women of European ancestry from the United States and Australia ., We identified IRF4 and SLC24A4 as loci highly associated with hair color , along with three other regions encompassing known pigmentation genes ., Further work is needed to identify the causal variants at these loci ., Improved understanding of the genetic determinants of human pigmentation may help identify the molecular mechanisms of pigmentation-associated conditions such as the tanning response and skin cancers . | public health and epidemiology/epidemiology, genetics and genomics | null |
journal.pgen.1004006 | 2,014 | Multi-tissue Analysis of Co-expression Networks by Higher-Order Generalized Singular Value Decomposition Identifies Functionally Coherent Transcriptional Modules | The increasingly cheaper and rapid accumulation of large -omics datasets across several experimental conditions has prompted generation of a wealth of data on biological networks ., This growth of network data now permits their large scale applications to biomedical research , including analysis of gene function , metabolic and signaling pathways , as well as disease-related or cell function-related networks 1 , 2 ., However , reconstructing and interpreting large biological networks , such as co-expression networks , protein-protein interaction networks or genetic networks , with different features ( e . g . , sparse or densely interconnected , etc . ) poses many challenges , advocating efficient and flexible methods for network inference and pattern discovery ., An important level of complexity in current network analysis regards its extension to multiple conditions , for instance different species 3 , cell-types 4 or disease states 5 , 6 ., For example , reconstruction of networks across multiple disease-states is becoming a useful approach for efficient drug-target discovery , as networks can inform the “biological context” ( e . g . , pathways , cellular processes ) where genes operate and therefore can help designing better therapeutic interventions 7 ., In genetic studies of complex diseases researchers increasingly focus on groups of highly interconnected genes within larger networks ( referred to as clusters , modules or subnetworks ) to elucidate specific cellular and molecular processes that might represent functional disease mechanisms and pathological pathways 8–10 ., While several computational tools for network analysis in single datasets or conditions are available , only few computationally efficient methods for genome-scale network analysis across multiple conditions have been developed to date ., These methods can be broadly classified into two main categories:, ( i ) methods to find the “difference” between networks across conditions or to pinpoint condition-specific networks 11–14 , or, ( ii ) methods to identify the common parts in networks across conditions 15–17 ., More recently , tensor-based computational frameworks 15 or probabilistic Markov blanket search algorithms 18 have been proposed to learn network structures across conditions ., However , these methods are either heavily influenced by the choice of input parameters ( e . g . , number of clusters , number of nodes within a cluster , cluster interconnectivity ) 15 or , being based on probabilistic graphical modelling , they become prohibitively slow for high number of conditions since they are trying to learn the structure of large graphs 18 ., Complementary to the above approaches , spectral methods , such as Singular Value Decomposition ( SVD ) , have been also proposed to investigate patterns of connectivity between nodes within a single network 19 , 20 or for comparing two networks 21 ., Generally , any network can be described as a graph , which is denoted as comprising a set of vertices or nodes together with a set of edges 22 ., The graph may be represented by a square , symmetric , real-valued matrix of size whose entries denote the relationship between the corresponding nodes ., In the affinity matrix , the element , called weight , represents the strength of connection between vertices and ., For instance , in gene regulatory ( or co-expression ) networks , the nodes might represent genes ( or mRNAs expression ) and edges represent the strength of gene-gene interactions ( or mRNAs co-expression ) ., Generalized Singular Value Decomposition ( GSVD ) can be used to identify sub-network structures and for comparative analysis of genomic datasets across two conditions 11 , 23 ., Given two matrices and 24 , 25 , their GSVD is given by ( 1 ) where and have orthonormal columns , is invertible , with , with ., The ratios are the generalized singular values of and ., In this setup , the common factor is informative of the cluster structure shared across the two data matrices ., Recently , a novel mathematical formulation , higher-order GSVD ( HO GSVD ) , which is constructed for more than two data matrices has been proposed 26 ., Under this framework , the matrices , each with full column rank ( i . e . , the maximum number of linearly independent column vectors of is ) , are decomposed as ( 2 ) where is composed of normalized left basis vectors , with and the latent factor matrix is composed of normalized right basis vectors ., The HO GSVD can be also derived in the special case of square , symmetric , full rank affinity matrices , , where each element represents the weight of the edge between node and in the th condition ., It has been previously employed to compare multiple datasets with identical column size in order to detect their common substructures of columns ( i . e . , observations ) 26 ., Yet , another useful application of the HO GSVD to genomics is to set it to discover gene networks across multiple conditions and pinpoint “common” and “differential” cluster structures ., In this paper , we build on the flexible HO GSVD mathematical framework and propose a new , parameter-free computational algorithm ( Cross-Conditions Cluster Detection or C3D ) for automatic detection of both similarity and dissimilarity clustering patterns in large weighted ( and unweighted ) networks across several conditions ( ) ., The original HO GSVD model has been employed for analysis of datasets that had varying number of genes ( ) , the same number of observations ( ) ( i . e . , arrays/time points in 26 ) across conditions and with ., As such , this illustrative application of the HO GSVD in genomics was aimed at the identification of common structures within the observations 26 ., Here , we built on the initial HO GSVD to extract sub-structures ( i . e . , common and differential clusters ) from genes across multiple conditions ( ) by applying the decomposition to the transposed expression matrix ., We show how this enables a more general application of the HO GSVD framework to genome-scale network analysis of genomic data ( e . g . , microarray , RNA-seq ) in multiple conditions ., Besides , a distinctive feature of our method is in its capability to take as an input either the raw expression matrices or co-expression matrices , allowing flexibility in the choice of the co-expression measures ( e . g . , Spearman , Kendall , mutual information , etc . ) ., Figure 1 illustrates the working principle of the C3D algorithm ., The input data for C3D can be provided into different formats to be used by the HO GSVD:, ( i ) the raw expression data matrices ( ) or, ( ii ) the co-expression data matrices ( ) ., In the former case , a first data initialization step is conducted where the input expression matrices , with the same number of genes are converted to co-expression matrices by scaling their variance to 1 and taking their quadratic form ., In the second step ( HO GSVD-based algorithm ) , an approximate HO GSVD is employed to identify a common basis , with representing the dimension of the GSVD common subspace , for the decomposition of the input datasets and identify the common and differential correlation structures ., The HO GSVD-based algorithm computes a square matrix , which is built on the arithmetic mean of all pairwise quotients where denotes the Moore-Penrose inverse of the co-expression matrix 24 ( see Methods section ) ., The first eigenvectors of ( according to the norm of the corresponding eigenvalues ) are then used to identify an approximate decomposition of the input co-expression matrices and form the decomposition basis ., Specifically , each selected column vector of is used to reorder the input data matrices such that candidate “common” ( or “differential” ) clusters can be identified ., In the third step ( cluster nodes selection and validation ) , we employ a mixture model approach to classify genes and assign them to each cluster based on a misclassification error rate ( MER ) ., Finally , we implemented an empirical cluster validation procedure to identify the conditions where clusters are present and assess the level of significance for clusters within each condition ., To demonstrate the increased power and benefits of our HO GSVD-based algorithm , we carried out an extensive simulation study and benchmarked C3D against commonly used methods that were designed to detect either common ( WGCNA 16 , 17 ) or differential network structures ( DiffCoEx 13 ) across multiple conditions ., We show that our approach has higher power and stability in detecting both common and differential co-expression clusters across all simulated conditions , while being two to seven fold less computationally intensive than alternative methods ., In contrast with alternative approaches that require specification of ad-hoc input parameters , the proposed method has the distinctive advantage of being parameter-free , which makes it a powerful tool for real data exploration and analysis ., To substantiate this claim , we applied C3D to publicly available transcriptomic datasets in rats and humans and identified several multi-tissue gene co-expression networks that were associated with specific functional processes relevant to phenotypic variation and disease ., We carried out a simulation study to compare our method with commonly used approaches for identification of “common” or “differential” clusters across multiple networks: ( 1 ) WGCNA and ( 2 ) DiffCoEx ., The WGCNA method for detection of common clusters across co-expression networks employs a “soft” threshold to assign a connection weight to each gene pair and extract densely connected gene clusters that are present in all conditions ., The DiffCoEx method follows a strategy similar to WGCNA but , instead , it focuses on detecting the differences in co-expression patterns ( “differential” clusters ) between multiple conditions ., Additional details on the specific parameterizations used in for WGCNA and DiffCoEx analyzes are reported in Text S1 ., To simulate a realistic example of gene expression data from multiple conditions that represent a typical “small large ” scenario , we draw inspiration from a publicly available multi-tissue microarray dataset consisting of genome-wide expression profiles from recombinant inbred rat strains in seven tissues 27 ., We simulated different types of clusters that are either detected in all conditions ( “common” clusters ) or are specific to a subset of conditions ( “differential” clusters ) , Figure, 2 . We considered dense clusters of variable sizes ( 100–500 nodes ) where each node is connected with all other nodes in the cluster with a given weight ( ) , which is defined as the Pearson correlation between expression profiles of genes and ., We simulated clusters with varying cluster densities ( 0 . 1 , 0 . 3 , 0 . 5 , 0 . 7 ) , which were defined as the average Pearson correlation between any pair of nodes within a cluster ., In addition to the simple case of a cluster common to all conditions and with the same size ( Cluster pattern 1 ) , we set out to evaluate the sensitivity of our and alternative approaches to detect clusters which are present only in a subset of conditions and that overlap partially across conditions ., This is more likely to be relevant for analysis of pathways and gene networks across tissues or during development , where varying gene-sets can exert their function only at specific developmental times or in specific cell-types ., To account for these more complex scenarios , we simulated “nested” ( Cluster pattern 2 ) and partially “overlapping” ( Cluster pattern 3 ) cluster structures ( Figure 2 ) ., Cluster pattern 2 and Cluster pattern 3 have an intersection part , defined by the nodes in common to all conditions , and a union part , defined by the nodes in common to all conditions plus the nodes present in individual conditions ., In summary , for each of the four cluster densities considered one dataset consisted of a and matrix in conditions , where each cluster type ( Clusters patterns 1–3 ) was simultaneously present in the data matrix ., To assess reliability of the results , for each of these data we generated 20 independent replicates , yielding a total of 560 simulated datasets ., Similarly , to evaluate how the number of available observations affects the methods performance we simulated datasets consisting of a and matrix in conditions ( 20 replicates , 560 datasets in total ) ., See Text S1 for additional details ., The True Positive Rate ( TPR ) and the False Positive Rate ( FPR ) are widely used as evaluation metrics for a classification model and can be used to quantitatively assess ( and compare ) methods performance 28 ., The TPR defines how many correct positive results ( simulated clusters genes within the called cluster ) occur among all results called positive in the analysis by a given method ., FPR , on the other hand , defines how many incorrect positive results occur among all results called positives ., Typically , a and the corresponding indicate a perfect classifier ( or a perfect method ) ., In our simulation study , the best cluster detection method would yield both high TPR and low FPR levels for different cluster types , sizes and densities ., For each simulated cluster type , Figure 3 shows the TP/FP rates for C3D , WGCNA and DiffCoEx methods as a function of the simulated cluster densities ., For C3D we controlled the ( local ) misclassification error ( i . e . , the probability to assign wrongly a gene to a cluster ) to be less than 0 . 05 or less than 0 . 2 , and required that each cluster is detected with , whereas for WGCNA and DiffCoEx we used two ( default ) parameterizations chosen according to the software guidelines ( see Methods section ) ., The C3D method outperformed WGCNA in the identification of clusters present in all conditions ( Cluster pattern 1 , Figure 3 ) , and showed to have consistently high TPR ( and very low FPR , ) irrespective of the simulated cluster density ., WGCNA performance varied considerably as a function of the simulated cluster density and , depending on the adopted parameterization , FPR levels were ( reaching 20% in one case ) , Figure, 3 . Furthermore , we observed large variations in WGCNA performance ( mostly in the TPR ) , which are indicated by the large standard deviations in TPRs calculated from the 20 replicated datasets ., For more complicated patterns ( “nested” and “overlapping” clusters ) , we compared C3D with WGCNA to detect the intersection part ( 100 nodes ) of common clusters ., Since WGCNA is designed to detect only those clusters shared across all conditions , for clusters present in a subset of conditions , we run WGCNA only in the set of conditions where the simulated clusters were present ., For Cluster patterns 2–3 , C3D and WGCNA performances were similar , reaching high TPR for detection of the intersection part of clusters with simulated ( Figure 3 ) ., However , C3D showed higher TPRs than WGCNA to detect clusters with low densities ( 0 . 1–0 . 3 ) , while controlling the FPR at low levels ( , Cluster pattern 2 intersection ) ., In the case of partially overlapping clusters present in a subset of conditions ( Cluster patterns 2–3 ) we compared C3D with DiffCoEx in respect of detecting the union part ( 500 nodes ) of “differential” clusters , and calculated TPR and FPR for detection of this cluster ( indicated with a black square at the top of Figure 3 ) ., We found that C3D outperformed DiffCoEx across the simulated scenarios ., In the case of the “nested” cluster structures that are present in 5 out of 7 conditions , C3D had consistently higher TPR levels than DiffCoEx , which showed comparable TPR levels only for detection of highly-dense clusters ( i . e . , , Cluster pattern 2 union , Figure 3 ) ., However , similarly to what observed for WGCNA method , in this case DiffCoEx showed large variability in its performance across the 20 replicated datasets ., The difference in performance between C3D and DiffCoEx was observed also in the more complicated case of partially overlapping cluster structures ( Cluster pattern 3 ) ., In this case , C3D showed consistently higher TPR than DiffCoEx that reached a maximum as compared with of C3D ., Both methods showed comparably low FPR ( ) for detection of the union part of Cluster patterns 2–3 ( Figure 3 ) ., Similarly to what observed for the simulated data with observations , C3D performed better than ( or as good as ) both WGCNA and DiffCoEx when benchmarked on simulated datasets with only observations ( Figure S1 ) ., As expected , all methods had lower TPRs associated with the detection of low-density clusters , however also with a small number of observations , C3D showed significantly better ( and more stable ) results than WGCNA and similar performance as compared with DiffCoEx ., Notably , for detection of “common” clusters present in all conditions ( Cluster pattern 1 ) , CD3 held high TPR levels ( and FPR ) whereas WGCNAs performance dropped significantly , reaching a maximum TPR ( Figure S1 ) ., These data show that C3D on balance performed better than WGCNA and DiffCoEx across all simulated scenarios ., We underline that while WGCNA and DiffCoEx methods are specifically designed to detect either common or differential clusters , respectively , here we showed that C3D was equally or more accurate than both methods in the detection of common and differential cluster structures ., We also highlight how C3D ability to detect correctly the simulated clusters was highly consistent across all runs on the replicated datasets , as shown by the small standard deviations of the mean TP and FP estimates ( Figure 3 ) ., In contrast , we observed that both WGCNA and DiffCoEx performances varied appreciably across the replicated simulations , often resulting in large standard deviations of the mean TP and FP estimates ., To better assess the reliability of the different methods we calculated the relative standard deviation of the TPR measured in all analyzed datasets ., In 560 simulated datasets of size , the C3D method had a median RSD of ( range 113 . 36 ) whereas WGCNA and DiffCoEx have median ( range 447 . 2 ) and median ( range 133 . 39 ) , respectively ., Similarly , in 560 datasets of size we estimated the following RSDs of TPR: 12 . 43 ( range 113 . 38 ) for C3D , 57 . 52 ( range 161 . 89 ) for WGCNA and 87 . 96 ( range 120 . 59 ) for DiffCoEx ., The large RSDs of TPR calculated from the WGCNA and DiffCoEx analyzes originated because these methods often detected the simulated cluster ( s ) only in small number of replicates ( e . g . , 2 out of 20 ) ., Besides , in a few cases the TP/FP rates of WGCNA and DiffCoEx were influenced by the adopted parameterization ( for instance , FPR in the WGCNA analysis of Cluster pattern 1 , Figure 3 ) , suggesting that different choices of the input parameters can affect the detection of clusters ( see Text S1 for additional details ) ., The C3D algorithm is built on the HO-GSVD framework and as such does not require the user to specify ad-hoc parameters to detect common or differential clusters ., In our implementation of the C3D algorithm the user can control the MER at a specified level before the cluster genes are empirically validated using a permutation-based procedure ( see Methods section ) ., In these simulation studies , we have used two different MERs ( 5% and 20% ) to inform a suitable choice of MER that maximizes true positive without inflating false positive rates ., On average , we observed a increase in the TPR when was adopted as compared with ., However , we found no significantly higher FPR , which were always across all simulated datasets , this suggesting that using the less stringent in real data analyzes is likely to increase the detection of true gene clusters , without increasing significantly false positives ., Finally , we used a standard desktop computer ( Mac Pro , GHz Quad-core Intel Xeon with 20 Gb RAM ) to evaluate the computational time required by C3D and compare it with WGCNA and DiffCoEx to analyze the simulated datasets ., While the run time of C3D scales exponentially with the number of genes in the input matrices or the number of conditions , our Matlab implementation of C3D is relatively fast and requires only 1 , 200s to analyze a gene co-expression matrix in conditions and 10s to analyze a gene co-expression matrix in conditions ( Figure S2 ) ., When compared with competing approaches , we assessed that to process simulated datasets of 1 , 000 and 10 , 000 genes ( with observations and conditions ) C3D requires significantly smaller CPU time than DiffCoEx ( up to 2 . 3 fold more CPU time ) and WGCNA ( up to 8 . 2 fold more CPU time ) , respectively ( Figure S2 ) ., To show how C3D provides a powerful , practical framework for real genome-scale analyzes and yields new biological insights into pathways and molecular networks , we report an application to two large multi-tissue gene expression datasets in rats and humans ., Transcriptional profiling was carried out by Affymetrix microarray in the rat and mRNA sequencing ( RNA-seq ) in humans , respectively ., The microarray dataset consisted of genome-wide expression profiles ( probe sets ) that were measured in seven tissues ( adrenal , aorta , fat , kidney , left ventricle , liver and skeletal muscle ) in a panel of recombinant inbred rat strains 29 , which is a well characterized model of hypertension , metabolic syndrome and cardiovascular disease 27 , 30 , 31 ., The RNA-seq datasets consisted of genome-wide transcriptomic data of human fetal neocortex , which have been generated to investigate the molecular mechanisms underlying differences in germinal zones of the developing human brain ., The human dataset consisted of expressed genes which were analyzed in four regions of the fetal neocortex ( ventricular zone ( VZ ) , inner subventricular zone ( ISVZ ) , outer subventricular zone ( OSVZ ) and cortical plate ( CP ) ) from six 13–16 weeks postconception human fetuses 32 ., In both rat and human analyzes , to identify common and differential clusters we extracted the top ten eigenvectors ( based on the modulus of the eigenvalues of the decomposition of ) as candidates which are then used as input for the cluster nodes selection and validation step of the C3D algorithm ( see Methods ) ., Building on the HO GSVD framework , we have developed a new algorithm ( C3D ) for efficient , parameter-free and automatic detection of co-expression clusters and networks in multiple conditions ., Our method is designed for analysis of weighted ( and unweighted ) networks ( input matrices ) across conditions , enabling applications to diverse data types and structures ., Although the original HO GSVD algorithm assumes the non-singularity of the co-expression matrix , by using the Moore-Penrose pseudo-inverse , our C3D algorithm can be applied to the non-invertible case ., We show that when an exact HO-GSVD of the input matrices exists ( as defined in ( 4 ) , see Methods ) , our HO GSVD is able to extract the right decomposition basis through the eigen-decomposition of , whereas it finds an approximate decomposition of the data in the absence of an exact solution ( Figure S4 ) ., In particular , our empirical simulations and real-case applications reveal that our approximate decomposition is able to capture both common and differential co-expression structures for a wide range of noise levels , suggesting that our algorithm can be useful for practical applications to genomic data ., Here , through the HO GSVD of large-scale genomic datasets we aimed to uncover the complex interactions between genes ( networks ) that can occur within or across multiple conditions ., One distinctive feature of our computational method is in the flexible and simultaneous identification of both “common” and “differential” sub-network structures across several conditions ., Selecting informative vectors of , we provide different orderings of to reveal candidate clusters that are important to all conditions or specific to a sub-set of conditions; then , we can distinguish the specific conditions where the clusters are present using a permutation-based approach ., This procedure allows to pinpoint automatically the specific conditions where the sub-network structures are present and , at the same time , to provide an empirical estimate of the statistical significance ( empirical P-value ) for each cluster identified ., In simulation studies , we demonstrated how C3D outperforms competing approaches in accuracy and reliability while being computationally less demanding ., We highlight how our method allowed accurate detection of clusters within complex structures ( i . e . , “common” , “nested” and “overlapping” networks ) by specifying only the desired level of statistical significance: misclassification error rate to assign genes to clusters and empirical P-value for cluster detection ., In contrast with other approaches , C3D does not need the user to specify ad-hoc parameters related to the expected number of clusters or cluster density 15 or necessary to determine the optimal height cut-off in the gene clustering tree 13 , 16 , 17 ., Typically , these unknown parameters need to be “finely tuned” on each dataset in order to obtain the best compromise between TP and FP for each cluster ( see Text S1 for additional details ) ., We also showed that the results obtained by two competing and widely-used methods ( WGCNA and DiffCoEx ) were less stable than those provided by C3D ., This was apparent in the significantly smaller relative standard deviations in TPR calculated across simulated datasets in the C3D analyzes as compared with WGCNA and DiffCoEx ., Since C3D utilised raw gene expression data matrices as input , the higher stability of C3D might be due to the reduced influence of the small number of observations on the stability of co-expression estimates , which can result in extreme patterns of correlation changes , corresponding to stable and fragile co-expression , as previously shown 62 ., The high stability in the results and the parameter-free “nature” of the HO GSVD approach make the C3D algorithm a powerful computational tool for real genomic data exploration and analysis ., To demonstrate this point , we reported an application of C3D to two large transcriptional datasets:, ( i ) microarray-based gene expression profiles in seven rat tissues and, ( ii ) RNA-seq-based gene expression analysis of germinal zones from human fetal neocortex ., In the rat analysis , we reported several functionally enriched co-expression clusters , including a previously identified inflammatory gene network driven by the IRF7 transcription factor that represents a gene expression signature of macrophages within complex tissues ., While this co-expression network was experimentally validated 27 it was not recovered by WGCNA , that surprisingly placed the IRF7 transcription factor and many regulated target genes in the group of “non-clustered” genes ., In addition , our C3D analyzes revealed novel gene co-expression networks in sub-sets of tissues ., For instance , we identified a network comprising Hsp and known cardiomyopathy genes , which suggested coordinated regulation of heat shock proteins genes in multiple tissues , and their potential functional role in cardiovascular disease 50 ., While this network was not recovered by either WGCNA or DiffCoEx analyzes , we were able to replicate this new finding using separate cardiac and liver gene expression datasets in humans ( Figure 4 ) ., In the study of human fetal neocortex we demonstrated previously undescribed co-expression between cell cycle and ECM-receptor interaction pathways and support their role in the proliferation and self-renewal of neural progenitors ., In addition , our analyzes highlighted that pathways central to later cognitive functions ( e . g . , calcium signaling , long-term potentiation , axon guidance ) are present at an early stage in the developing human brain 61 , which was not previously appreciated ., These studies illustrated how our method can be effectively applied to leverage the vast stream of genome-scale transcriptional data that has risen exponentially over the last years , promising to aid the fine-scale characterization of both context-specific and systems-level networks and pathways ., In this step we assume the input data are non-square matrices , where the rows represent the observations and the columns indicate genes ., The number of genes must be the same across datasets while the number of observations can differ ., We first log transform the data and subtract for each gene its average gene expression to avoid capturing differences in average gene expression across conditions ., We then calculate the co-expression matrices corresponding to each condition ., Each represents the covariance matrix of the data in condition ., As in classic principal component analysis , the columns of can be scaled to unit variance to work on the correlation matrices rather than the covariance ., Alternatively , our algorithm can directly take any co-expression matrix as input ., This feature of our algorithm allows to extract common and differential clusters from matrices based on different co-expression measures , including robust correlation ( e . g . Spearman , Kendall ) and non linear metrics such as mutual information 63 ., Similarly to classic SVD , each observation from the input data can be characterized by its expression profile and represented by a data point in a dimensional space ., The observations from all datasets are contained in a subspace of dimension , which thereafter is referred to as the HO GSVD subspace ., Here , we aim at finding directions in the HO GSVD subspace that either capture the variability in gene expression that is common to all conditions ( common factors ) or that is specific to a subset of conditions ( differential factors ) ., Inspired by 26 we developed a general algorithm that allows computation of an approximate solution to the HO GSVD problem in the non full column rank case ., In the HO GSVD , are decomposed into where , is a diagonal matrix with elements for and contain the right basis vectors of the HO GSVD subspace where ., The right basis vectors allow to identify set of genes ( clusters ) with similar co-expression patterns , that are either specific to a subset of conditions or common to all conditions ., Here we explain the derivation of our HO GSVD-based algorithm in the general case of non-square matrices ., The derivation and discussion of the special cases ( square , symmetric matrices with full rank and square , symmetric matrices with full rank ) is reported in Text S1 ., In the most general case , we define the right basis vectors as the solution of the eigen-decomposition problem of the matrix ( 3 ) where is the arithmetic mean of all the pairwise quotients and denotes the Moore-Penrose inverse of the co-expression matrix 24 ., Here the Moore-Penrose inverse is used as a substitute of since the invertibility of is not guaranteed when , which is the typical scenario in genomics ., We now assume there is an approximate HO GSVD where is composed of orthonormal left basis vectors and ., In this case , for all we have ( 4 ) and its Moore-Penrose inverse is given by ( 5 ) Therefore we have ( 6 ) since is full row rank ., Hence we can rewrite as follows ( 7 ) When there exists a common subspace of dimension , with basis vectors , for which the decomposition of the co-expression matrices ( 4 ) is exact , equation ( 7 ) becomes an equality and the eigenvectors of will lead to the exact basis of the common subspace ., In HO GSVD applications to genomics data , can be as large as the total number of observations ( i . e . , ) , and an exact common decomposition of the co-expression matrices might not be possible ., In this case the eigenvectors of do not provide an exact decomposition of the subspace ., Moreover , is not guaranteed to be non-defective and have a full set of real eigenvalues and eigenvectors ., However , even in the absence of an exact common d | Introduction, Results, Discussion, Methods | Recent high-throughput efforts such as ENCODE have generated a large body of genome-scale transcriptional data in multiple conditions ( e . g . , cell-types and disease states ) ., Leveraging these data is especially important for network-based approaches to human disease , for instance to identify coherent transcriptional modules ( subnetworks ) that can inform functional disease mechanisms and pathological pathways ., Yet , genome-scale network analysis across conditions is significantly hampered by the paucity of robust and computationally-efficient methods ., Building on the Higher-Order Generalized Singular Value Decomposition , we introduce a new algorithmic approach for efficient , parameter-free and reproducible identification of network-modules simultaneously across multiple conditions ., Our method can accommodate weighted ( and unweighted ) networks of any size and can similarly use co-expression or raw gene expression input data , without hinging upon the definition and stability of the correlation used to assess gene co-expression ., In simulation studies , we demonstrated distinctive advantages of our method over existing methods , which was able to recover accurately both common and condition-specific network-modules without entailing ad-hoc input parameters as required by other approaches ., We applied our method to genome-scale and multi-tissue transcriptomic datasets from rats ( microarray-based ) and humans ( mRNA-sequencing-based ) and identified several common and tissue-specific subnetworks with functional significance , which were not detected by other methods ., In humans we recapitulated the crosstalk between cell-cycle progression and cell-extracellular matrix interactions processes in ventricular zones during neocortex expansion and further , we uncovered pathways related to development of later cognitive functions in the cortical plate of the developing brain which were previously unappreciated ., Analyses of seven rat tissues identified a multi-tissue subnetwork of co-expressed heat shock protein ( Hsp ) and cardiomyopathy genes ( Bag3 , Cryab , Kras , Emd , Plec ) , which was significantly replicated using separate failing heart and liver gene expression datasets in humans , thus revealing a conserved functional role for Hsp genes in cardiovascular disease . | Complex biological interactions and processes can be modelled as networks , for instance metabolic pathways or protein-protein interactions ., The growing availability of large high-throughput data in several experimental conditions now permits the full-scale analysis of biological interactions and processes ., However , no reliable and computationally efficient methods for simultaneous analysis of multiple large-scale interaction datasets ( networks ) have been developed to date ., To overcome this shortcoming , we have developed a new computational framework that is parameter-free , computationally efficient and highly reliable ., We showed how these distinctive properties make it a useful tool for real genomic data exploration and analyses ., Indeed , in extensive simulation studies and real-data analyses we have demonstrated that our method outperformed existing approaches in terms of efficiency and , most importantly , reproducibility of the results ., Beyond the computational advantages , we illustrated how our method can be effectively applied to leverage the vast stream of genome-scale transcriptional data that has risen exponentially over the last years ., In contrast with existing approaches , using our method we were able to identify and replicate multi-tissue gene co-expression networks that were associated with specific functional processes relevant to phenotypic variation and disease in rats and humans . | biotechnology, algorithms, systems biology, computer applications, computer science, mathematics, algebra, genetics, applied mathematics, computing methods, biology, genomics, computational biology, numerical analysis | null |
journal.pcbi.1002732 | 2,012 | Linkers of Cell Polarity and Cell Cycle Regulation in the Fission Yeast Protein Interaction Network | The eukaryotic cell cycle is one of the most important and evolutionary conserved processes of cells 1 , 2 ., The cell cycle integrates signals from multiple pathways to control tissue growth and homeostasis in multicellular organisms , as well as reproduction and proliferation in single cell organisms 3 ., To ensure cell integrity , the cell cycle regulates and is regulated by other key processes such as DNA replication , cytokinesis and cell growth 4–9 ., Disruption of the regulation between the cell cycle and other cellular processes can cause a myriad of cellular pathologies including defects in cell shape , abnormal cell growth and aneuploidy , potentially leading to cancer 10 ., With the accumulation of data from high-throughput biology as well as the generalisation of manually curated online databases , we now can mine existing biological networks to make experimentally verifiable predictions about system-wide properties of genes and gene products ., In this work , we present a new method to search for proteins that serve as linkers between distinct functional sub-networks ., Because of the well-characterized interactions between the cell cycle and other processes in the fission yeast Schizosaccharomyces pombe , we focus our analysis on this organism , where these processes have not yet been investigated yet by protein interaction network analysis methods ., The fission yeast - a rod-shaped unicellular eukaryote - is ideally suited to study the relationship between cell cycle and cell polarity regulation , as its highly polarized growth pattern is tightly correlated with cell cycle progression 7 , 11 ., After cytokinesis , newborn S . pombe cells resume growth in G1 in a monopolar fashion from their ‘old end’ - the cell end that existed prior to division - and later in early G2 activate growth at their ‘new end’ derived from the site of septation , an event termed new-end take-off or NETO 12 ., Bipolar growth then continues through G2 until cells reach a critical size , after which cells enter M phase again ., At that point cells stop growing 13 , mitosis takes place and each cell divides by growing a septum in its middle ., Daughter cells resume their cyclic pattern of growth at the ends and division at the middle , a pattern that relies on the cytoskeleton of actin and microtubules and on diverse polarity-regulating proteins ( ‘polarity factors’ ) ., Cytokinesis , polarity , and the cell cycle have been extensively studied in fission yeast –using both experiments and mathematical modelling 14–18 ., The insights gained from studies in fission yeast often carry over to higher eukaryotes , as the molecular machinery controlling those processes has been highly conserved throughout evolution 1 , 19 , 20 ., Several proteins have been identified that play important roles connecting these processes in fission yeast ., For example , the polarized growth-regulating DYRK kinase Pom1 21 was recently shown to form a spatial gradient that is used by the cell cycle machinery to sense the length of the cell 17 , 22 , 23 ., Another link was observed between the morphogenesis-related NDR kinase network ( MOR ) and the septation initiation network ( SIN ) 24 ., MOR is important for the localization of actin patches to sites of polarized growth , while SIN is responsible for triggering cytokinesis ., It was discovered that SIN inhibits the MOR pathway , through inhibition of the Orb6 activator Nak1 ., MOR itself also interferes with SIN , and this antagonism is required for proper progression through the cell cycle 25 , 26 ., Furthermore , a similar antagonism between the MOR and SIN pathways has also been observed in higher eukaryotes 27 , 28 ., The NETO transition from monopolar to bipolar growth and the switch from polarized growth to actin ring-mediated cell septation are also controlled by the cell cycle 13 , thus the cell cycle machinery enforces a major control on both polarized growth and cytokinesis ., Although many polarity or cytokinesis regulators contain potential phosphorylation sites for the cell cycle-regulating Cyclin-Dependent Kinases 29 ( CDK ) , the molecular details of these couplings are not well known ., In the other direction , if either polarized cell growth or cytokinesis is inhibited , both can send signals to stop the cell cycle 30 , 31 , further underlining that these three functional modules are highly interlinked ., To tackle the interplay between different cellular processes , we utilized a network theory approach ., Hitherto , network based approaches have only been used in a limited number of organisms , due to the paucity of genome-wide interaction data available for most species ., Recently , however , improvements in automatic experimental annotation , literature mining 32 , machine learning 33 and orthology annotations 34 , are allowing the use of network approaches in a wider range of organisms ., For example , ‘meta databases’ such as STRING 35 , 36 , benchmark information from multiple sources and provide for each possible interaction a confidence score that reflects the likelihood of a set of proteins of actually interacting ., Here , we take advantage of such developments and build on the efforts of the fission yeast community in annotating protein functions 37–39 , to establish a new method to identify proteins linking diverse cellular processes , based on integrating Gene Ontology ( GO ) 40 , 41 and Protein-Protein Interaction ( PPI ) data together with network theory based measures ., Network-based approaches in biology have been used in the past to identify community structures , study lethality , identify specific regulatory circuits and study hierarchical organization 42 ., In particular , the nature of large scale protein-protein interaction networks has recently been under considerable debate with different groups disagreeing about the modularity of networks , as well as the properties of the nodes responsible for bringing together different modules 43–46 ., In this work , we sidestep the difficult problem of identifying hierarchical modules in a large , genome-wide network and focus instead on a method to identify proteins that link different cellular processes ., To do this , we use the highly characterized sub-genomic network consisting of proteins regulating the cell cycle , cytokinesis , and polarized cell growth in fission yeast ., We propose a new network measure , termed ‘linkerity’ , and use it to predict a novel role for a number of proteins as key bridges between these biological processes ., We constructed the fission yeast protein-interaction network using data from STRING 35 , 36 and BioGRID 47 ., By applying a cutoff on the confidence score from STRING , we can reject interaction pairs for which there is a limited amount of evidence ( see Materials and Methods for details on data in STRING ) and use the remaining edges to construct a non-directed and non-weighted network ., We then examined the effects of increasing the cutoff in STRING confidence scores in both the genome-wide interaction dataset of fission yeast and that of the better characterized budding yeast Saccharomyces cerevisiae on the network topology ., Increasing the cutoff decreased the amount of nodes ( Figure 1A ) and the edge density ( Figure 1B ) in the largest component ( the connected component in the network containing the highest number of edges and nodes ) of both the fission and budding yeast networks ( Tables S1 , S2 ) ., This decrease was less sharp in budding yeast compared to fission yeast due to the extensive amount of genome-wide interaction experiments carried out in the former , increasing the amount of high-confidence interactions ., Interestingly , in the ‘core’ sub-network consisting of proteins involved in cell cycle regulation , polarity and cytokinesis ( Figure 2 for fission yeast and Figure S1 for budding yeast ) , the drop off in the number of nodes and edges was far less significant in both yeasts , suggesting that interaction data for the core fission yeast network tends to be more reliable than interaction data for the rest of the network ( Figure 1 , red stars versus red dots , also Tables S1 , S2 , S3 , S4 ) ., As a more stringent test , we constructed networks for both organisms using only data from BioGRID 47 ., BioGRID is a database that only contains data from manually annotated experiments ( distinguishing between experiments that show direct physical interaction and genetic interactions ) ., Networks built using the BioGRID physical interaction data also show that the core networks of fission yeast and budding yeast are relatively dense , while the fission yeast organism-wide network is rather sparse ( Figure 1 ) ., Even with the relatively high coverage of the core ( regulation of cell cycle , cytokinesis , polarity ) network in fission yeast , it is important to note that fission yeast lacks any genome-wide protein-protein interaction experiments , and as such , several of the interactions predicted by STRING are based on indirect evidence such as genetic interactions , inference from homology , or literature mining 35 , 36 ., As no analysis of the fission yeast network has been previously published , we performed a few checks to verify that our network construction procedure was giving sensible results , and that the data for fission yeast available in STRING was of sufficiently high quality ., As a first check , we sought to replicate a number of analyses previously performed with budding yeast ( Table 1 ) ., At a cutoff of 0 . 7 ( defined by STRING as a ‘high confidence’ threshold ) , the genome-wide fission yeast network has 2770 nodes with at least one connection and 20432 edges compared to 5477 nodes and 105429 edges found in budding yeast , although they have approximately similar number of proteins ., We calculated the degree distribution for the nodes in the network , and observed that , as previously described for numerous other complex networks 48 , the fission yeast PPI network has a scale-free distribution ( Figure S2 ) ., We also repeated analyses done in numerous other studies examining the relationship between network measures and gene deletion lethality 43–45 ., As reported for budding yeast , we observed that degree ( the number of interactions with other proteins ) is the best predictor network measure of gene deletion-induced lethality in fission yeast , and that ratio of these essential genes among hubs ( the top 20% of proteins by degree ) is even higher in fission yeast than in budding yeast ( see Text S1 ) ., Since there is no high-throughput genome-wide interaction data available for fission yeast , we tested the possibility that highly investigated proteins might have more interactions ., To check this , we tested to see whether the number of abstracts in PubMed discussing a particular protein was correlated with the degree of that protein in the network ., The Pearson correlation between the number of PubMed abstracts citing a protein and its degree in the network was 0 . 13 for budding yeast ( p-value<10−19 ) and 0 . 14 for fission yeast ( p-value<10−13 ) ( details in Tables S1 , S2 ) , suggesting there is no fission yeast-specific bias for proteins with large amounts of publications in STRING networks ., However , a large amount of evidence for the fission yeast interactions in STRING is obtained from homology , and specifically from interactions of homologues proteins in budding yeast ., As essential genes are more likely to be conserved 49 , 50 and STRING is more likely to identify homology between highly conserved genes , it is possible that this might introduce a subtle bias making essential genes appear to be more highly connected in virtue of their higher conservation ., This is consistent with the observation that a very high percentage of hubs in fission yeast appear to be essential ( Text S1 ) ., The sub-network of all proteins regulating cell cycle , cytokinesis and polarized growth , henceforth , the ‘core’ network ( see Materials and Methods for definitions of exact GO terms used ) in fission yeast contains 550 proteins: 384 of those are associated with regulation of cell cycle , 155 with cytokinesis and 139 with polarity ., Using a cutoff of 0 . 7 in STRING , 429 of the total 550 proteins are connected to the largest connected component of the core network ., Most of the proteins not in the network have no known interactions , and the second largest connected component contains only 4 proteins , thus we focus only on the interaction network of the largest connected component ., There are a high number of proteins with multiple functions in the network ( Figure 2A ) , 16 of them ( Alp4 , Cdc15 , Gsk3 , Lsk1 , Mor2 , Orb6 , Pab1 , Pmo25 , Pom1 , Ppb1 , Ras1 , Scd1 , Shk1 , Sid2 , Tea1 , Wsp1 ) are important for all three cellular processes and 77 have dual functions ., The ratio of multifunctional proteins is quite similar to the ratio in the analogous core budding yeast network ( Figure S1 ) ., Interestingly the budding yeast core network contains less nodes than the fission yeast core network ( although it is more densely connected ) , this could be a consequence of the extensive studies of cytokinesis 19 , cell cycle 51 and cell polarity 13 and their careful annotation in fission yeast 37–39 , but it also reflects the loss of some of the conserved eukaryotic cell cycle genes from budding yeast 29 , 52 ., The core interaction network contains several interactions between proteins that do not share a GO annotation; however the majority of links ( 91% ) are between proteins which share at least one functional annotation among those under consideration ( regulation of cell cycle , cytokinesis , and polarity ) ( Figure 2B ) ., To probe this , we examined the relationship between the functional annotation of a node and that of its interaction partners ., In fission yeast , any protein with a given functional annotation was 11 times ( 1 . 9 would be expected randomly , see Figure S3A ) more likely to interact with another protein with the same functional annotation than with another protein with different functional annotations ( for the budding yeast core network , this ratio was 4 . 5 vs . 1 . 06 expected , see Figure S3B ) ., Since fission yeast has more proteins that belong to all three categories ( 16 in fission yeast versus 6 in budding yeast ) , we tested to see whether this observed functional modularity was due to their presence ., We removed all proteins belonging to all three categories from both networks and repeated the analysis ., This did not significantly alter the results as the ratios remained after the removal ( 10 . 38 times more likely for fission yeast and 4 . 16 for budding yeast ) suggesting that the functional modularity observed in fission yeast is not caused by the presence of highly connected proteins with multiple annotations , but rather that the fission yeast network is characterized by strong connections between local communities that share functional annotations ., It is however important to note that the GO categories ‘regulation of cell cycle’ and ‘cytokinesis’ are partially overlapping ., In particular ‘regulation of cell cycle cytokinesis’ is a child term of both ‘regulation of cell cycle’ and ‘cytokinesis’ ., Even when taking this overlap into account in the analysis , we still observe a high degree of functional modularity in the core networks of both fission and budding yeast ( not shown ) ., We further analyzed this effect using a community detection algorithm , which identifies local communities in a network and allows their overlap – as we have nodes with multiple annotations ., We applied the k-clique propagation algorithm 53 , 54 and examined the communities generated by the method with k\u200a=\u200a4 ., While the communities generated by the algorithm do not exactly match the functional annotations , we find that the cliques generated by the algorithm are primarily formed by proteins that share functional annotations ( Figure 3A , B ) ., Upon closer examination , the few proteins that do not share a functional annotation with the other members of a clique seem to have related roles: for example , in the 5th clique on Figure 3B , the lone ‘non-polarity’ protein is Rgf3 , which was shown to play an important cell-wall remodeling role downstream of Rho1 , one of the key regulators of polarity 55 , 56 ( consult Table S5 for all clique members ) ., To systematically study proteins linking different cellular processes , we next used a network-based approach aiming to identify proteins that function as ‘linkers’ between different functional categories ( Figure 4A ) ., To do so , we constructed protein-protein interaction networks consisting only of proteins with one of the investigated functional annotations ( cell cycle , cytokinesis or polarity regulation ) ., We then calculated the betweenness centrality score for every node in each of these networks and in the merged core network ., Betweenness Centrality ( BC ) measures how often a node is found in the shortest path between pairs of other nodes in the network; intuitively , it can be thought of as a measure of how central a node is in a network ., If a node has a low centrality score it is localized at the fringe of a network , while if it has a high score it is localized near the centre ., Next we ranked the proteins based on their BC score ( in case of a tie , these proteins got their average rank ) ., To ensure that this ranking method is robust even in the presence of imperfect interaction data certainly missing important links , we randomly added 10% extra edges to all the networks 1000 times , and recalculated the ranking of all proteins at each iteration ( Figures S4 ) ., While the exact ranking of proteins is not very robust to addition of extra edges , if we examine all the proteins in the top 20% , we can observe that most fluctuate out of the top 20% only very rarely , and that we nearly never observe a protein in the top 10% drop out of the top 20% ., It is also reassuring that the top of the rankings starts with expected key regulators of each function: the polarity landmark Tea1 57–59 , the actin-regulating Rho GTPase Cdc42 60 , 61 and actin ( Act1 ) all came on the top of the polarity list ., At the same time Cdc2 , Wee1 and Cdc25 62 are on the top of the cell cycle list ( and also on the top of the core list ) and the SIN scaffold Cdc11 63 and the CDK counteracting , SIN activator phosphatase Clp1 64 , 65 are leading the cytokinesis ranking ( Figure S4 and Table S3 ) ., In the next step we compared the betweenness centrality rank of every protein in a sub-network to its relative rank in the core network ., Only proteins that were originally in the sub-network were considered during this ranking based on scores they got for their position in the core network ., We then calculated the ratio of the relative rank in the core network and the rank in the sub-network ., We termed this calculated value ‘linkerity’ , as this value is high for proteins that are found at the fringe of the network of proteins controlling a given cellular process , but central when considered in the context of a bigger network ( Figure 4A ) : ( 1 ) Proteins with high linkerity , we hypothesized , are likely to play a crucial role to function as linkers between different cellular processes ., Specifically , we focused on the relationship of the polarity network to the rest of the core network to clarify how the cell cycle and the cytokinesis machinery control the temporal changes in the localization of polarized growth zones ( top of Table 2 , consult Table S3 for the rest of the list ) ., Here , we show the top 10 proteins with the highest linkerity scores ., These proteins became far more central when the polarity sub-network was embedded into the core network ., Most of these proteins have GO annotations for multiple processes ( among the annotations under consideration ) , thus their linking capacity is not that surprising ., Novel linkers of polarity regulation could be those that were not associated with cytokinesis or cell cycle control but gained a high linkerity score in our analysis ., The formin For3 66 , the AMP-activated , Snf1-like protein kinase Ssp2 67 , 68 , the RNB-like protein Sts5 69 and the MRG family protein Alp13 70 are examples of proteins that match this ., For3 is a well-characterized regulator of Tea1 to Cdc42 signalling 71 , 72 , the other three are less well characterized ., The Rho GTPase Rho4 73 might be also an interesting linker candidate as it has established roles in polarity and cytokinesis regulation , but its exact function is not well characterized and it has no association to cell cycle regulation ., Despite this , Rho4 has a central position in the core network that contains 75% cell cycle proteins ( Figure 2A ) , furthermore its expression is cell cycle regulated 74 ., The highest linkerity proteins from the cytokinesis and cell cycle regulation networks also contain a number of proteins which are also associated with polarity regulation ( Table 2 ) ., Scd1 , Pom1 and Tea1 are on the top of the cell cycle linkerity list and Pmk1 75 , Shk1 and Tea1 lead the cytokinesis list after Bgs1 , which is essential for cell wall synthesis 76 , but has no polarity related GO annotation ., These are on the edge of the cell cycle regulation or cytokinesis network but became central when they are merged with the polarity network , thus these can be also considered as linkers ., As above for BC scores , we analysed the robustness of linkerity in the presence of imperfect network interaction data: we added or removed 10% of the edges from the core network at random or following a preferential attachment model and calculated linkerity scores for all proteins ., Figure 4B reports the average and standard deviation from 500 random networks with 10% extra edge ( other cases in Figure S5 ) for the top linkerity polarity proteins ., Importantly the top 10 of the unperturbed list ( Table 2 ) can be found in the top 16 of the list after 10% possible missing links were considered ( Figure 4B ) ., As discussed above , in both fission yeast and budding yeast , we observe a high degree of functional modularity , i . e . proteins tend to interact with proteins that share their functional role ., Since linker proteins play a special role in bringing together different cellular processes , we examined whether proteins with high linkerity interacted with proteins with different functional roles at a higher rate than low linkerity proteins ., For all the proteins of the core network we calculated the number of its interactors ( network neighbours ) with cell cycle , cytokinesis and polarity annotations ( Table S3 ) ., Then for every protein in each functional category ( Figure 2 ) we calculated the ratio of the number of its interactions with proteins with the two other functional annotations to the number of its interactions with proteins with the same functional annotation ., We observed that high linkerity is significantly correlated with having a high ratio of heterogeneously annotated neighbours across all functional categories in both yeasts , suggesting that linker proteins do play an important role in bridging proteins from different functional groups ( see Text S2 for details ) ., Among predicted linker proteins we focused on Sts5 , which is known to genetically interact with Ssp2 69 , which itself is likely to be linked with the cell cycle machinery as ssp2Δ cells cannot start mitosis when nutrient-starved 77 ., Sts5 is an orthologue of budding yeast SSD1 78 and therefore a candidate translational repressor ., It is reported to control actin localisation in interphase and sts5Δ was shown to be compensated by mutations in Ssp2 ., Furthermore , Sts5 mRNA levels were shown to oscillate 74 , 79 ., To examine the interplay between Sts5 and the cell cycle , we tagged the endogenous protein with a triple GFP tag and visualized its localization together with that of mCh-Atb2 ( Alpha tubulin 2 ) , which labels microtubules and hence served as a cell cycle stage marker ., In interphase cells , Sts5 had a mostly diffuse cytoplasmic localization , however during mitosis it appeared to localize in dotted , cytoplasmic bodies ( Figure 5A ) ., The number of Sts5 dots increased throughout mitosis and peaked coinciding with the assembly of the Post Anaphase Array ( PPA ) of microtubules ( Figure 5B ) ., Time-lapse movies of mitotic cells also confirmed that the number of cytoplasmic dots increased until the formation of the PAA and sharply dropped to zero as cells entered interphase ( Figure S6 ) ., Previous studies of Sts5 69 showed that it was required for correct cell growth and actin patch localization during interphase ., Taken together with our results , this suggests that the cell cycle controls Sts5 activity by gradually sequestering it in cytoplasmic bodies during mitosis ., In this work , we have carried out the first network analysis based , large-scale identification of proteins linking various cellular processes in the fission yeast protein-protein interaction network ., Although data for fission yeast mostly comes from manually annotated experiments , literature mining and computational inference , the network displays features comparable to those observed in other organisms ., We have shown that the relationship between lethality and different network measures holds in fission yeast , and that network based approaches can give meaningful and interesting results even in organisms lacking high-throughput interaction experiments ., Our analysis of the core network of all proteins regulating cell cycle , cytokinesis , and polarized growth revealed a striking degree of functional modularity , which we have found to be highly robust to the deletion of key nodes in the network ., This functional modularity was also observed when examining the communities detected by a clique propagation algorithm ., Detected communities had very low heterogeneity between the functional annotations of member proteins ., We investigated this modularity further by using a network approach to identify linker proteins bridging different functional categories ., We propose a new network measure , linkerity , which is the ratio of the ranking by betweennness centrality measures of all the nodes belonging to a given sub-network considered in the sub-network alone and considered in the context of a larger network ( Figure 4A ) ., This new network measure does not appear to show strong correlation with other existing network measures ( Text S3 ) ., Due to the non-linear distribution of betweenness centrality measures in real systems 48 , it might be necessary to normalize this linkerity measure in case linkers between large sub-networks are investigated ., We tested this concept on the connections of the polarized cell growth regulatory network to the cytokinesis and cell cycle networks of fission yeast cells ., These are highly characterized and strongly interacting networks and the connection between these processes is of high importance in other organisms 7 , 13 , 80–82 ., We confirmed that many of the highest linkerity scoring proteins in the polarity network were already known to play important roles in multiple processes ., Among these the F-BAR protein Cdc15 provide good validation as it was already shown to play a role in switching from polarized growth to cytokinetic-actin ring formation in mitosis 83 ., Similarly Skb1 84 and Cdr1 17 , 23 were shown to serve as links between cell cycle and cell polarity ., All these proteins shifted from a low ranking in the polarity network to a high rank in the core network ( Table 2 ) , and thus their role in polarity regulation might come from the pleotropic behavior of these proteins or from their active role in connecting polarized growth regulation to cell cycle and cytokinesis ., We also discovered that the proteins with high linkerity tend to interact with a more diverse set of proteins than those with low linkerity ., This suggests that high linkerity proteins might play a pleiotropic role by linking together different functional processes 85 , 86 ., Sts5 had the second highest ranking in the polarity network among the top ten linkerity proteins ( after actin , Act1 that is also essential for cytokinesis ) ., Sts5 is known to play an important role in controlling the localization of the actin machinery to cell ends during interphase , although Sts5 is localized in the cytoplasm 69 ., We have shown that Sts5 is localized in cytoplasmic dots during mitosis , but diffuse during interphase , implying that its localization is cell cycle regulated ., Growing tip localized polarity proteins change their localization when cells enter mitosis 13 , 87 , but it is not expected from a cytoplasmic protein to localize into clusters in a cell cycle dependent manner ., The overall level of Sts5 protein slightly increases upon entry to mitosis ( Figure S6 ) , but its activity reaches its lowest level as its accumulation into cytoplasmic dots reaches a peak ., This suggests that the cell cycle controls polarity by sequestering Sts5 in and out of cytoplasmic bodies , and the triggered release and sequestration function as switches between polarized cell growth and cytokinesis ., The exact nature of those cytoplasmic bodies is still unclear , however the budding yeast Sts5 homologue SSD1 was shown to localize to P-bodies 88 , the cytoplasmic centers of mRNA degradation ., Interestingly , like Sts5 , Ssp2 and the stress pathway kinase Wis4 are also localized into cytoplasmic dots 89 and it was proposed that the stress pathway and Sts5 might act in opposing manner on cell polarity 68 ., It will be important in the future to investigate if these proteins co-localize in the observed cytoplasmic dots and how these are exactly controlled by the cell cycle ., Sts5 was previously shown to genetically interact with members of the stress pathway 69 ., A number of other kinases associated with stress response ( such as Sty1 , Skb1 , Orb6 , Pmk1 , Mkh1 ) have been shown to have defects in NETO 84 , 89 and many of these appear highly ranked in our linkerity lists ( Table 2 ) ., Furthermore , the cell end-localized polarity factor Tea4 was also shown to interact with the stress pathway 90 ., These make the stress pathway a particularly intriguing target for further analysis in the search for proteins linking cell cycle and polarity , as it may play a special role as a pleiotropy integrator of both internal and external cellular signals in response to different stimuli in fission yeast and also in higher eukaryotes 91 , 92 ., The linkerity analysis of cytokinesis and cell cycle regulatory proteins ( bottom parts of Table, 2 ) also give some interesting predictions ., For instance the high linkerity of the transcription factor Cdc10 93 in the cell cycle network suggests its role controlling the transcription of important polarity and cytokinesis genes , especially with key regulators , such as Cdc15 , Scd2 , Sts5 , Rho4 and Sid2 having periodic transcriptional profile 74 , 79 ., While we believe that the method presented here can be applied to other organisms and cellular processes to find linker proteins , different model organisms offer unique advantages and challenges ., In this study , we took advantage of the extensive annotation of proteins by the fission yeast community to define discrete sub-networks , bypassing the very difficult problems involved in defining meaningful ‘communities’ using purely network based approaches 46 , 54 , 94 ., While this approach has its advantages , it is important to be aware of any partial overlaps between the used GO terms due to the presence of common child terms ., The amount of overlap between child terms is also not consistent across multiple organisms , requiring special care when doing comparisons that involve multiple organisms ( for example , the “regulation of cell cycle cytokinesis” is a child term of both “regulation of cell cycle” and “cytokinesis” and it contains 47 p | Introduction, Results, Discussion, Materials and Methods | The study of gene and protein interaction networks has improved our understanding of the multiple , systemic levels of regulation found in eukaryotic and prokaryotic organisms ., Here we carry out a large-scale analysis of the protein-protein interaction ( PPI ) network of fission yeast ( Schizosaccharomyces pombe ) and establish a method to identify ‘linker’ proteins that bridge diverse cellular processes - integrating Gene Ontology and PPI data with network theory measures ., We test the method on a highly characterized subset of the genome consisting of proteins controlling the cell cycle , cell polarity and cytokinesis and identify proteins likely to play a key role in controlling the temporal changes in the localization of the polarity machinery ., Experimental inspection of one such factor , the polarity-regulating RNB protein Sts5 , confirms the prediction that it has a cell cycle dependent regulation ., Detailed bibliographic inspection of other predicted ‘linkers’ also confirms the predictive power of the method ., As the method is robust to network perturbations and can successfully predict linker proteins , it provides a powerful tool to study the interplay between different cellular processes . | Analysis of protein interaction networks has been of use as a means to grapple with the complexity of the interactome of biological organisms ., So far , network based approaches have only been used in a limited number of organisms due to the lack of high-throughput experiments ., In this study , we investigate by graph theoretical network analysis approaches the protein-protein interaction network of fission yeast , and present a new network measure , linkerity , that predicts the ability of certain proteins to function as bridges between diverse cellular processes ., We apply this linkerity measure to a highly conserved and coupled subset of the fission yeast network , consisting of the proteins that regulate cell cycle , polarized cell growth , and cell division ., In depth literature analysis confirms that several proteins identified as linkers of cell polarity regulation are indeed also associated with cell cycle and/or cell division control ., Similarly , experimental testing confirms that a mostly uncharacterized polarity regulator identified by the method as an important linker is regulated by the cell cycle , as predicted . | cellular structures, protein interactions, signaling networks, microbiology, model organisms, cell division, cell growth, cytokinesis, cytoskeleton, regulatory networks, schizosaccharomyces pombe, biology, proteomics, systems biology, biochemistry, cell biology, yeast and fungal models, computational biology, molecular cell biology, genetics and genomics | null |
journal.ppat.1005957 | 2,016 | Dissemination and Mechanism for the MCR-1 Colistin Resistance | The polymyxins ( polymyxin E ( colistin ) and polymyxin B ) are a family of cationic polypeptide antibiotics with a lipophilic fatty acyl side chain 1 , 2 ., The initial binding of polymyxins bacterial surface mainly depends on the electrostatic interaction between the positively-charged polymyxin and the negatively-charged phosphate group of lipid A on lipopolysaccharide ( LPS ) localized on the outer leaflet of the bacterial outer membrane 2 ., Following its diffusion from the outer membrane across the periplasm , polymyxin intercalates into the inner membrane and forms pores , which in turn results in bacterial lysis 2 ., Although they belong to an old generation of antibiotics , polymyxins represent the last line of defense against lethal infections by gram-negative pathogens with pan-drug resistance 3 ., Unfortunately , certain species of the Enterobacteriaceae like K . pneumoniae 3 have been recently showing an appreciable resistance to colistin ., Indeed , colistin resistance ( i . e . , inefficient binding of polymyxins to the lipid A moiety of lipopolysaccharide ) is mainly due to the 4’-phosphoethanolamine ( PEA ) modification of the lipid A on the LPS 4 , 5 ., This type of chemical modification on the bacterial lipid A can be attributed to either the chromosome-encoded machinery in K . pneumoniae 6 or the plasmid-transferred mobilized colistin resistance ( MCR-1 ) gene in certain species of Enterobacteriaceae like E . coli 7 ., For the former , two sets of bacterial two-component systems ( pmrAB 8 plus phoPQ 6 ) and the regulator mgrB 6 are implicated , in which the lipid A of LPS is chemically modified and thereafter exhibits reduced affinity to polymyxin 7 ., The latter represents an unique mechanism for bacterial colistin resistance in that the mcr-1 gene product , annotated as a member of a family of phosphoethanolamine transferases , catalyzes the modification of lipid A moiety on bacterial LPS ( Fig 1 ) 2 , 7 ., To the best of our knowledge , the natural occurrence of the mcr-1 gene has been traced to no less than five species: E . coli 7 , 9 , 10 , Salmonella enteric 11 , K . pneumonia 7 , Enterobacter aerogenes 12 and E . cloacae 12 ( of note , it was also experimentally spread/transmitted from E . coli to Pseudomonas aeruginosa by conjugation 7 , 13 . ) ., Also , the range of host reservoirs with potential to carry the mcr-1-harbouring enterobacteria extends from poultry/livestock ( chickens 11 , pigs 7 , 11 , 14–16 , dogs 17 , and cattle 11 ) to humans 10 , and published data from January-April 2016 suggests that the mcr-1 gene has been disseminated into no less than 18 countries 10 ., To a certain degree , the global spread of the mcr-1 gene might be related to a food-chain based dissemination pathway , which was shown by Zhu’s group 11 ., Thus , they observed the paralleled existence of mcr-1 in meat/food samples and in the healthy human microbiome 11 ., Worryingly , the MCR-1 colistin resistance gene was strikingly shown to coexist with other multiple-drug resistance genes ( i . e , carbapenem 18 and extended-spectrum β-lactam 16 , 19–21 ) , highlighting the possibility that micro-organisms with pan-drug resistances are emerging 22 ., For instance , a variant of the notorious NDM-1 was detected to coexist with MCR-1 in the Enterobacteriaceae ( NDM-5 in K . pneumoniae 23 and NDM-9 in a chicken meat isolate of E . coli 24 ) ., So far , most of the studies in this field focused on epidemiological investigations , which is in part due to the relatively limited availability of the genomic information ., Nevertheless , the mechanism for transfer , origin , and biochemical analysis of the diversified plasmid-borne MCR-1 colistin resistance remains poorly understood , and these questions are addressed here , in aiming to close the missing knowledge gap ., The recent emergence of colistin resistance may be attributed to MCR-1-mediated PEA addition of lipid A moiety at the 4’-phosphate group ( Fig 1A ) , a component of bacterial LPS on the outer layer of outer-membrane for Gram-negative bacteria like E . coli ( Fig 1B ) 7 ., Consistent with scenarios seen in Neisseria 2 and E . coli 7 , our MS result suggested that the peak of lipid A with mass of 1797 . 4 is present in the colistin-susceptible E . coli MG1655 ( S1A Fig ) , and one more peak of PPEA-4’-lipid A with mass of 1919 . 8 appears upon the arabinose-inducing expression of the MCR-1 enzyme ( S1B Fig ) ., It validated that the essence of MCR-1-catalyzed enzymatic reaction is the addition of PPEA ( mass: 123 Au ) to lipid A ( mass: 1797 . 4–1797 . 6 ) ., Very recently , we successfully isolated mcr-1-harbouring plasmids from the colistin-resistant E . coli strains obtained from the gut microbiota of clinically diarrheal patients 10 ., Here we subjected the mcr-1-positive plasmids to genome sequencing by next generation desktop MiSeq sequencer ( Illumina ) ., A pool of 350-bp paired-end reads we produced , were assembled with GS De Novo Assembler into two long contigs ., We then checked the assembly of our plasmids ( pE15004 and pE15017 ) through integrating raw data , PCR assays , and Sanger sequencing ., The plasmid pE15004 was assembled correctly , while a ~2 . 2 kb fragment was missing in the original assembly of pE15017 ., Consequently , complete genomes of these two clinical mcr-1-positive plasmids ( pE15004 , in Fig 2A and pE15017 in Figs 2B and 3A–3D ) were acquired ., The mcr-1-harbouring plasmid pE15004 was 33 . 309 kb in length with a GC content of 41 . 8% ., This plasmid contained 51 predicted ORFs , among which 11 were associated with the formation of type IV pilus ( Fig 2A ) ., The backbone of pE15004 was closely-related to the pir-type E . coli plasmids pSAM7 from cattle in the United Kingdom and pJIE143 from human in Australia ( S2 Fig ) , both of which are narrow-host-range IncX4-type plasmids 25 , 26 ., Further comparative analysis indicated that plasmid pE15004 was nearly identical to IncX4 mcr-1-harboring E . coli plasmids pICBEC72Hmcr ( Accession no . : CP015977 ) isolated from Brazil and pESTMCR ( Accession no . : KU743383 ) from Estonia , K . pneumoniae plasmids pMCR1_Incx4 ( Accession no . : KU761327 ) from China ( Fig 3E ) 27 and pMCR1 . 2-IT ( Accession no . : KX236309 ) identified in Italy 28 ( Table 1 ) ., Another mcr-1-bearing plasmid pE15017 ( 65 . 375 kb ) we sequenced contained 91 ORFs ( Fig 2B ) , sharing nearly all its sequences with the first-identified mcr-1-harbouring IncI2-type plasmid pHNSHP45 7 ( Fig 3 and S2 Fig ) ., In comparison with the prototype mcr-1-positive plasmid pHNSHP45 , the upstream insertion sequence ISApl1 flanked mcr-1 was consistently missing in both pE15004 and pE15017 ( Fig 3E ) as well as in other recently-reported mcr-1-containing plasmids , like pKH457-3-BE 22 ., Besides the mcr-1 gene ( Fig 3A and 3C ) , our PCR assays combined with Sanger sequencing determined that pE15017 carries an extended-spectrum β lactamase ( ESBL ) gene blaCTX-M-55 ( Fig 3A and 3B ) ., Thus pE15017 represents an ESBL and MCR-1-coproducing plasmid ( Fig 3A , 3D and 3E ) ., Previously , the co-occurrence of ESBL and MCR-1 had been found on the IncHI2-type plasmids from E . coli and Salmonella enterica 20 , 29 ., Similar to the other mcr-1-carrying plasmid , pA31-12 ( Fig 3D ) 30 , pE15017 might be an additive example of IncI2-type plasmid with above two antibiotic resistance determinants ., Interestingly , four base pair ( AACA , 1612–1615 ) is consistently missed in the ISEcp1-blaCTX-M-55-orf477 operon ( 3090 bp ) in both pE15017 and pA31-12 , in relative to the counterpart ( 3094 bp ) in pHN122-1 ( Fig 3D ) 31 ., In fact , the ISEcp1-blaCTX-M-55 transposition unit flanked by DR was also recently observed in the Salmonella plasmid pSCS23 ( KU934209 ) as shown in Fig 3 29 ., Taken together , these results indicate that plasmid pE15004 is an additive member of the mcr-1-carrying plasmids , while pE15017 together with other recently identified plasmids such as pmcr1_IncI2 ( Accession no . : KU761326 ) 23 and pBA76-MCR-1 ( Accession no . : KX013540 ) , are recent variants of plasmid pHNSHP45 ( Figs 2 , 3E and S2 Fig ) ., Furthermore , the mcr-1 gene has been found carried by plasmids belonging to at least 5 different incompatibility groups ( Table 1 ) , verifying a trend of diversification 10 ., Though it is currently only found in Enterobacteriaceae , its dissemination to broad host range plasmids and subsequent spread to a broad bacterial host range cannot be excluded ., Moreover , comparative analysis showed that nearly identical mcr-1-containing plasmids were discovered in different countries , suggesting that , besides the possibility that the dissemination of the mcr-1 gene was captured independently from a common ancestor 11 , the direct spread of bacteria harboring the same plasmid is not impossible ., Although Liu and coworkers determined that the expression of plasmid-borne mcr-1 confers colistin resistance to certain species of the Enterobacteriaceae family 7 , functional details of the MCR-1 protein are poorly understood ., Here , we attempt to address this issue ., Philius Transmembrane Prediction Server ( http://www . yeastrc . org/philius/pages/philius/runPhilius . jsp ) suggested that the MCR-1 protein is an integral membrane protein with five trans-membrane regions ( S3A Fig ) ., Similar to the LptA ( EptA ) of Neisseria , the multiple sequence alignments indicated that the MCR-1 protein also belongs to a family of phosphoethanolamine lipid A ( PEA ) transferases with putative conserved sites ( E246 , T285 , H395 , D465 and H466 , in S3B Fig ) required for its catalytic activity , i . e . , the addition of PEA to lipid A from phosphatidylethanolamine ( Fig 1 ) ., Because the fact that the nascent LPS in cytoplasm is flipped by the ABC transporter MsbA into periplasm 32 and the covalent modification of the lipid A component on LPS occurs in periplasm 1 , it is speculated that the trans-membrane regions ensures the correct anchoring of the MCR-1 enzyme to the periplasmic face of the cytoplasmic membrane attached to the catalytic domain of PEA transferase ., While , experimental evidence for this hypothesis is lacking , we aimed to address them using the integrative approaches ranging from protein biochemistry , bioinformatics and structural biology to bacterial genetics ., We over-expressed the membrane protein MCR-1 and purified it to homogeneity ( S4A Fig ) and confirmed this by Western blot using an anti-6x-His primary antibody ( S4B Fig ) ., MS-based identification further confirmed the identity of the recombinant MCR-1 trans-membrane protein ( S4C Fig ) ., To further gain structural insights into the biochemical mechanism of MCR-1 , we applied both protein engineering and structure-guided mutagenesis ., In particular , the arabinose-inducible expression system pBAD24/MG1655 was also utilized to probe the above concerns in vivo ., Given the fact that, i ) the chemical modification phosphate group of lipid A at 1 or 4’-position impair its binding to polymyxins 2 ,, ii ) the newly-synthesized LPS is translocated by the MsbA lipid flippase into periplasm from cytoplasm 1 , 32;, iii ) bacterial periplasm is the only place where the moiety of lipid A on LPS is covalently modified with either 4-amino-arabinose or phosphoethanolamine 2 , it is prerequisite that the enzyme modifier including MCR-1 should be localized in bacterial periplasm ., Thus , we are extremely interested in determining physiological role of the trans-membrane region in MCR-1 function ., Firstly , MG1655 with/without the empty vector pBAD24 ( the negative control ) fails to grow on the LBA plates with above 2 mg/L of colistin , whereas the positive control , MG1655 with the arabinose-induced expression of mcr-1 , can grow well on the solid media with 16–32 mg/L of colistin ( Table 2 ) ., In contrast , pBAD24-facilitated expression of Neisseria lptA conferred the colistin-susceptible MG1655 strain with an ability to grow on the LBA plates with 8–16 mg/L colistin ( Table 2 ) ., Though the Neisseria lptA encodes the colistin resistance at an appreciable level , the amplitude of drug resistance is less than that of the MCR-1 ( Table 2 ) ., Subsequently , we engineered a deletion mutant of the mcr-1 gene ( Δtm ) whose protein product lacks the N-terminal trans-membrane region to further evaluate its role in vivo ., Similar to the scenario with the negative control , we found that the E . coli strain with the araninose-induced expression of the mcr-1 mutant ( Δtm ) cannot grow on the LBA plates with over 2 mg/L of colistin ( Table 2 ) , validating the importance of the transmembrane region in the MCR-1-mediated colistin resistance ., Thus it may be concluded that the catalytic activity for PEA transferase depends on its location of MCR-1 on bacterial inner-membrane ., Using the Neisseria Lipo-oligosaccharide Phosphoethanolamine Transferase A ( LptA ) as structural template ( PDB: 4KAV ) 2 , structural modeling by Swiss-Model program allow us to visualize the architecture of PEA transferase domain of the membrane-bound MCR-1 enzyme ( Fig 4A ) ., Extensive analyses of structural docking together with sequence alignments allow us to hypothesize that the following five residues ( E246 , T285 , H395 , D465 , and H466 ) are critical for the substrate binding of MCR-1 , and in turn determines the MCR-1-encoded colistin resistance ( Fig 4B ) ., Driven by this speculation , we used site-directed PCR mutagenesis to create the following point mutations ( E246A , T285A , H395A , D465A , and H466A ) ., In contrast to the positive control carrying the wild-type mcr-1 gene ( Fig 4C ) , none of the MG1655 strains expressing the mcr-1 point mutants were observed to grow significantly on the condition of above 2 . 0 mg/L of colistin , which is almost identical to that of negative control ( Fig 4C and Table 2 ) ., This represents in vivo evidence that the five residues are essential for the function of MCR-1 ., A BLASTp search for MCR1 and Neisseria gonorrhoeae LptA ( EptA ) returned a large set of divergently related sequences ( annotated as PE transferases , Sulfatases or membrane proteins ., Detailed comparisons of alignment methods applied to divergently related sequences have produced low-accuracy results with sequence identities below 30% 33 ., We have thus limited our search scope to 30% ., To determine a phylogenetic profile of MCR-1 , Multiple sequence alignment of the dataset was performed by MUSCLE ( http://www . ebi . ac . uk/Tools/msa/muscle ) 34 using default parameters and the quality of the alignment was evaluated using Guidance ( http://guidance . tau . ac . il ) 35 ., Consequently , we retrieved 32 candidate proteins that returned hits with >30% identity ., Maximum Likelihood ( ML ) phylogenic trees were reconstructed by using the LG amino acid substitution model with gamma distribution and invariant sites selected using MEGA version 6 36 ., To ensure appreciable reliability , the results we obtained were validated by 1000 bootstrap replicates ., Intriguingly , the reconstruction of a maximum phylogeny tree using 32 unique proteins selected from the BLASTp search allowed us to clearly observe two distinctive clades: one containing a family of PEA transferases including MCR-1 and Neisseria LptA ( Fig 5 ) and the other containing an array of putative sulfatases ( Fig 5 ) ., Also , the members of the PEA transferase family are divided into two apparent subclades , one of which features MCR-1 , and the other one comprising Neisseria LptA ( Fig 5 ) ., The chromosomally-encoded LptA from Neisseria species is most-closely clustered with putative integral membrane proteins found in other pathogenic Ɣ-proteobacteria , whereas the plasmid-borne MCR-1 on the other hand is very close to PEA transferases from the colistin-producing bacteria , esp . the Paenibacillus species ( Fig 5 ) 37–39 ., Despite the fact that MCR-1 and LptA share very low sequence identity to each other and fall into two separate subclades within the tree ( Fig 5 ) , they still remain functionally-equivalent ( Table 2 ) ., The phylogenetic tree here indicates a divergent evolutionary pattern between the LptA/MCR-1 integral membrane proteins and other putative sulfatases ., A domain analysis of MCR-1 revealed distinct trans-membrane helices followed by a sulfatase domain ., Sulfatases that catalyze the hydrolysis of a sulfate group are present in all three domains of life and constitute a heterogenic group of enzymes 40 , 41 ., Due to similarity in size between a sulfate and a phosphate group , one can easily imagine why PEA transferases share core catalytic features with sulfatases ., In fact , closely-related sulfotransferases from Mycobacterium transfer a sulfate group into the glycolpeptidolipids ( GPL ) , the equivalent of the LPS in gram-negative bacteria 42 ., Due to lack of sufficient sequence data and experimental validation , it is hard to trace the ancestry of MCR-1 to its chromosomal origins ., Given the fact that removal of the trans-membrane region from the MCR-1 protein damages its function of MCR-1-mediated colistin resistance ( Table 2 ) , one can speculate that acquisition of a trans-membrane domain could have easily enabled these PEA transferases to correctly localize in the inner membrane and to eventually target a variety of substrates with different implications ranging from cationic antimicrobial peptide resistance in the case of lipid A modification , to changes in motility when FlgG , a flagella rod protein , is modified ., Given the fact that phylogenetic tree places the MCR-1 protein very close to the PEA transferases from the Paenibacillus family , the known producers of polymyxins , but in a different sub-clade than Neisseria LptA ( Fig 5 ) , it raises the possibility that, 1 ) the cousins of Paenibacillus might be highly relevant to its origin of the MCR-1;, 2 ) a potentially parallel evolutionary path is implicated for the two genes ( mcr-1 and lptA ) under similar environmental selection pressures , e . g . , the massive use of colistin as a veterinary medicine ., The data we present represents a first comprehensive glimpse of mechanisms for diversified plasmid transfer , evolutionary origin , and catalytic reaction of the MCR-1-mediated colistin resistance ., The discovery of new mcr-1-harbouring plasmids ( pE15004 and pE15017 ) adds new knowledge into the newly-emerging field of MCR-1 and colistin resistance , furthering our understanding of the diversity in the dissemination of the mcr-1 gene 11 ., Unlike the paradigm mcr-1-positive plasmid pHNSHP45 that is isolated from a swine E . coli in southern China , the two plasmids we reported are extracted from clinical E . coli isolates of diarrhea patients ., Further functional definition of plasmid genomes delineated that, 1 ) the plasmid pE15004 is a IncX4 plasmid of around 33kb long 25 , 26 , differing from the IncI2-type plasmid pHNSHP45 of about 64kb in length 7 ( Figs 2 , 3 and S2 Fig ) ;, 2 ) the other plasmid pE15017 ( ~65kb ) seemed to be a recent variant of pHNSHP45 ( ~64kb ) in that an insertion sequence ISApl1 of around 1 kb in front of mcr-1 gene of pHNSHP45 is absent in pE15017 ( Fig 3E ) and an ESBL encoding gene was captured ( Fig 3A , 3B and 3D ) ., In agreement with proposal by Petrillo and coauthors 43 , the deletion of this ISApl1insertion sequence might be the relic of the mcr-1 dissemination ., It was reported that the mcr-1 colistin resistance gene is present in a multidrug-resistant plasmid 22 or coexists with other resistance genes like extended spectrum β-lactamase 16 , 19–21 , and even the notorious NDM-1 44 and its variants ( NDM-5 23 and NDM-9 24 ) ., We also found that the pE15017 is an ESBL and MCR-1 coproducing plasmid similar to pA31-12 , a recently-isolated plasmid from China 30 ., These facts imply that multi-drug , even pan-drug , resistant bacteria with colistin resistance will eventually evolve , a fact that deserves close attention ., By contrast , the pE15017 plasmid carries both ESBL and MCR-1 ., Consistent with our recent observations with swine lung microbiota 45 , the results highlight the differences amongst the mcr-1-carrying plasmid reservoirs in human gut microbiota 10 ., We experimentally validated that the expression of Neisseria LptA augments colistin resistance of E . coli ( Table 2 ) , despite being weaker than MCR-1 ( Table 2 ) , suggesting the possibility of various catalytic aspects and differing evolutionary paths for the two genes ( lptA and mcr-1 ) ., To address this concern , we conducted phylogenetic analyses and found they are placed into two neighboring sub-clades of the PEA transferase family ( Fig 5 ) , giving a strong implication of parallel evolutionary paths for the two genes ( mcr-1 and lptA ) ., Additionally , the functional impairment of the MCR-1 colistin resistance by the removal of the trans-membrane regions demonstrates that membrane anchoring of the soluble catalytic domain ( PEA transferase ) is essential its function ( Table 2 ) ., In particular , we also determined the requirement for the five motif-forming residues for MCR-1 function ( Table 2 and Fig 4 ) , which might facilitate the binding of this enzyme to its cofactor of zinc ions ., The mechanistic insights we obtained definitely extended our knowledge on MCR-1/colistin resistance , and might provide a molecular basis for the development of inhibitors/drugs of small molecule via bypassing the MCR-1-mediated colistin resistance in the post-antibiotics era ., Generally , the resistance to colistin is correlated with the decrement in affinity of the lipid A group of lipopolysaccharide to the polymyxin antibiotics ., Unlike the chromosome-encoded mechanism with the involvement of a two-component systems ( pmrAB 8 and phoPQ 6 ) and the regulator mgrB 6 , the plasmid-borne MCR-1 represents an newly-emerging machinery for colistin resistance in which the modification of lipid A is catalyzed by the MCR-1 enzyme , giving the reduced affinity to polymyxin ( Fig, 1 ) 7 ., It seems likely that the current situation of MCR-1 colistin resistance worldwide has been over-underestimated since almost 30 countries have been identified to have the mcr-1 gene present in the past several months 5 , 10 ., Given the fact that, i ) colistin is the last line of refuge amongst therapeutics against lethal infections by multidrug-resistant Gram-negative pathogens 3 , 46; and, ii ) the extensive consumption of colistin as a veterinary medicine in the poultry/swine production worldwide functions as a strong selective pressure which then imposes a risky burden on food safety and public health , it is urgently needed to reconsider appropriate use of colistin in veterinary/human medicine and restrict global dissemination of the mcr-1 colistin resistance gene transferred by diversified plasmids ., In summary , our findings provide a functional glimpse of plasmid transfer , evolutional origin , and catalysis mechanism for the MCR-1 colistin resistance ., The clinical E . coli strains from gut microbiota of diarrhea patients ( kindly provided by Shenzhen Centre for Diseases Control , China 10 ) were grown in the liquid Luria-Bertani ( LB ) media for the isolation of mcr-1-positive plasmids ., The two genetically-modified strains DH5α and BL21 ( DE3 ) were separately applied for gene cloning and protein expression ( S1 Table ) ., The colistin-susceptible strain of E . coli MG1655 was used for functional assays for the mcr-1 gene and/or its mutants ( S1 Table ) ., The solid LB agar plates supplemented with appropriate antibiotics were applied to either screen possible positive clones for the presence of the mcr-1 gene or determine the minimum inhibitory concentration of colistin by expression of MCR-1 ., The plasmids were isolated routinely from E . coli strains using an alkaline lysis method ., Using specific primers ( S2 Table ) , PCR screening was performed for the presence of mcr-1 gene in the colistin-resistant strains ., The full coding sequence of mcr-1 was then cloned in pET28 ( a ) via the two cuts ( BamHI plus XhoI ) , giving the recombinant plasmid pET28::mcr-1 ( S1 Table ) ., Both the wild type of mcr-1 and its deletion mutant , mcr-1 ( Δtm ) were directly inserted into the two cuts ( EcoRI and SalI ) of an arabinose-inducible expression vector pBAD24 , giving the plasmids of pBAD24::mcr-1 and pBAD24::mcr-1 ( Δtm ) , respectively ( S1 Table ) ., Similarly , the recombinant plasmid pBAD24::lptA was constructed through cloning of the Neisseria LptA-encoding gene into pBAD24 ( S1 Table ) ., Using the pBAD24::mcr-1 plasmid as the template , The experiments of site-directed PCR mutagenesis were conducted as we earlier described 47 ., All the acquired plasmids were verified by PCR assays and direct DNA sequencing ., The minimal inhibitory concentration ( MIC ) of colistin was determined using liquid broth dilution test as recommended by EUCAST with Cation-adjusted Mueller-Hinton Broth ., And survival ability of E . coli expressing different protein was determined as follows: mid-log phase cultures diluted appropriately were spotted on LBA plates supplemented with colistin at varied level ( ranging from 0 , 0 . 5 , 1 . 0 , 2 . 0 , 4 . 0 , 8 . 0 , 16 . 0 to 32 . 0 mg/L ) , and maintained at 37°C overnight ., In the assays for colistin resistance/tolerance , the colistin-susceptible strain MG1655 acted as the negative control , the MG1655 strain carrying the empty vector , pBAD24 , referred to blank control , and the MG1655 strain with pBAD24::mcr-1 is the positive control ., All the other strains that expressed either Neisseria LptA , the trans-membrane deletion mutant of mcr-1 ( Δtm ) , or one of and five mutants of mcr-1 with single point mutation , were tested for their different ability of colistin resistance ., 0 . 2% arabinose was added into the LBA plates to induce the expression of lptA/mcr-1 ( and/or its mutants ) ., Ultrapure LPS was extracted using the hot phenol method as described by two different groups 2 , 7 with minor modifications ., Briefly , overnight E . coli cultures ( ~10 ml ) collected by centrifugation were washed with 5 ml of 50% cold acetone before re-suspending in 0 . 55 ml of water at 70°C ., It was then mixed with 0 . 45ml of phenol ( pre-warmed to 70°C ) by vigorous vortexing ., This mixture was incubated at 70°C for 30 min before spinning at 16 , 000x g for 15 min to collect the aqueous phase ., 1 . 3 ml of cold absolute ethanol , 6 . 7 μl of 3 M sodium-acetate and water were added till a final volume of 1 . 9 ml ., This was incubated at -80°C for 15mins to precipitate the crude LPS ., The resultant crude LPS from the aqueous phase was dialyzed against de-ionized water using the aqueous phase 1000 MWCO dialysis tubing ., After dialysis , the samples were freeze-dried , and re-suspended in 55 μl wash solution ( 20 mM Tris-HCl ( pH 8 . 0 ) , 2 mM MgCl2 , DNase I ( 100 μl of 7 mg/ml ) and RNase A ( 100 μl of 17 mg/ml ) ) ., The mixture was maintained for 3 h at 37°C prior to adding 5μl of Proteinase K and further incubating it at 56°C for 1hr ., An equal amount of phenol was mixed with the mixture and centrifuged for 30mins at 16 , 000x g to collect the aqueous phase ., 193 μl of 50 mM TRIS , 7 μl of 3M sodium acetate ( pH 5 . 2 ) and 3 volumes of cold ethanol were added to the aqueous phase and incubated at -80°C for 15mins to precipitate the LPS ., The precipitated LPS was collected by centrifugation at 16 , 000x g for 15 mins and re-suspended in 50 μl water ., Finally , Lipid A samples were assayed with electrospray ionisation mass spectrometry ( ESI-MS ) as Liu et al . 7 reported ., The mcr-1-positive plasmids that met the requirement of quality control were subjected to library preparation prior to the whole genome sequencing ., The next-generation Illumina MiSeq sequencing was conducted as per protocols recommended by the manufacturer , generating a pool of 350-bp paired-end reads ., The draft assembly of plasmids was performed with GS De Novo Assembler to give two long contigs ., PCR and Sanger sequencing were then performed to verify and correct the contigs ., As a result , full genomes of the two plasmids of clinical origins ( pE15004 and pE15017 ) were produced ., The plasmid sequences were annotated by RAST , and the genome maps were drawn with the Circos program ., Comparative genomics of plasmids were carried out with the tools , Glimmer and BLAST , to probe the potential origin/mechanism for transfer of the mcr-1-carring plasmids ., To identify sequences homologous to MCR-1 , a BLASTp search was performed using the amino acid sequence of MCR-1 and Neisseria gonorrhoeae LptA ( formerly named EptA ) as a query ., In order to avoid hits from very closely related species , Escherichia coli and uncultured environmental samples were excluded from the search and the max target sequences acquired were 500 ., The top unique protein sequences were selected and submitted to the web-based program , Guidance ( http://guidance . tau . ac . il ) 35 , to evaluate the quality of alignment and to identify potential regions and sequences reducing the quality of alignment ., Multiple sequence alignment was performed using MUSCLE with default parameters ( http://www . ebi . ac . uk/Tools/msa/muscle/ ) 34 ., The alignment was also manually evaluated and adjustments were made as necessary ., The best amino acid substitution model to be used for reconstructing a tree was identified using the Models function in MEGA version 6 36 ., The model with the least score ( LG with G+I ) was used to reconstruct Maximum Likelihood trees while treating gaps/missing data as partial deletions ., Results were validated using 1000 bootstrap replicates ., Philius Transmembrane Prediction Server ( http://www . yeastrc . org/philius/pages/philius/runPhilius . jsp ) was applied to probe the topological structure of the MCR-1 protein ., The protein sequences of MCR-1 and the related proteins were subjected to the program of Clustal Omega ( http://www . ebi . ac . uk/Tools/msa/clustalo/ ) , and the final output of the multiple sequence alignments was given processed by the program ESPript 2 . 2 http://espript . ibcp . fr/ESPript/cgi-bin/ESPript . cgi ) 48 ., Structural modeling for the PEA-lipid A transferase domain of MCR-1 was processed by Swiss-Model program , using the Neisseria Lipo-oligosaccharide Phosphoethanolamine Transferase A ( LptA ) as structural template ( PDB: 4KAV ) 2 , the resultant result in ribbon structure was given via PyMol software ., Plasmid typing was performed with the help of PlasmidFinder-1 . 3 server ( https://cge . cbs . dtu . dk/services/PlasmidFinder/ ) ., The genome sequences of the two plasmids ( pE15004 & pE15017 ) were deposited into the GenBank database with the accession no ., , KX772777 and KX772778 , respectively . | Introduction, Results, Discussion, Materials and Methods | Polymyxins are the last line of defense against lethal infections caused by multidrug resistant Gram-negative pathogens ., Very recently , the use of polymyxins has been greatly challenged by the emergence of the plasmid-borne mobile colistin resistance gene ( mcr-1 ) ., However , the mechanistic aspects of the MCR-1 colistin resistance are still poorly understood ., Here we report the comparative genomics of two new mcr-1-harbouring plasmids isolated from the human gut microbiota , highlighting the diversity in plasmid transfer of the mcr-1 gene ., Further genetic dissection delineated that both the trans-membrane region and a substrate-binding motif are required for the MCR-1-mediated colistin resistance ., The soluble form of the membrane protein MCR-1 was successfully prepared and verified ., Phylogenetic analyses revealed that MCR-1 is highly homologous to its counterpart PEA lipid A transferase in Paenibacili , a known producer of polymyxins ., The fact that the plasmid-borne MCR-1 is placed in a subclade neighboring the chromosome-encoded colistin-resistant Neisseria LptA ( EptA ) potentially implies parallel evolutionary paths for the two genes ., In conclusion , our finding provids a first glimpse of mechanism for the MCR-1-mediated colistin resistance . | Colistin is an ultimate line of refuge against fatal infections by multidrug-resistant Gram-negative pathogens ., The plasmid-mediated transfer of the mobile colistin resistance gene ( mcr-1 ) represents a novel mechanism for antibacterial drug resistance , and also poses new threats to public health ., However , the mechanistic aspects of the MCR-1 colistin resistance are not fully understood ., Here we report comparative genomics of two new mcr-1-harbouring plasmids isolated from the human gut microbiota ., Genetic studies determined that both the transmembrane region and a substrate-binding motif are essential for its function ., Phylogenetic analyses revealed that MCR-1 is highly homologous to the PEA lipid A transferase in Paenibacillus , a known producer of polymyxins ., The fact that the plasmid-borne MCR-1 is placed in a subclade neighboring the chromosome-encoded colistin-resistant Neisseria LptA potentially implies parallel evolutionary paths for the two genes ., Our results reveal mechanistic insights into the MCR-1-mediated colistin resistance . | sequencing techniques, antimicrobials, medicine and health sciences, enzymes, split-decomposition method, drugs, enzymology, microbiology, multiple alignment calculation, antibiotics, molecular biology techniques, pharmacology, plants, bacteria, research and analysis methods, legumes, sequence analysis, polymyxins, lipids, antimicrobial resistance, neisseria, sequence alignment, proteins, transferases, peas, molecular biology, biochemistry, computational techniques, microbial control, biology and life sciences, organisms | null |
journal.pcbi.0030231 | 2,007 | An Evolutionary-Network Model Reveals Stratified Interactions in the V3 Loop of the HIV-1 Envelope | The human immunodeficiency virus type 1 ( HIV-1 ) possesses a highly variable envelope comprising the glycoproteins gp120 and gp41 , which mediate the binding and entry of the virus into a host cell ., The viral envelope is also a potent antigen for neutralizing antibodies 1–4 and cytotoxic and helper T lymphocytes 5–7 , which is manifested as extensive sequence divergence in the env gene 8 , 9 ., Consequently , HIV-1 is required to maintain a functioning envelope while accumulating a sufficient number of mutations in env to escape the adaptive immune response ., This conflict can be surmounted by the evolving virus populations through selection for combinations of substitutions that exploit structural or functional interactions among residues in the envelope glycoproteins 10 ., A structural interaction occurs between residues that cooperate in the formation and stabilization of secondary or tertiary protein structures ., On the other hand , a functional interaction is a statistical association that arises indirectly between residues that participate in the same protein function , e . g . , key residues in a conformational binding site or glycosylation motif ., Redundancy that arises from such interactions allows residues to be replaced by other combinations while conserving the overall phenotype ., This phenomenon , known as compensatory mutation , features prominently in HIV-1 evolution 11–13 and is pervasive across all levels of biological diversity 14 ., The detection of interactions among residues in rapidly evolving viral proteins such as the HIV-1 envelope is an important and unresolved problem ., First of all , the failure to account for such interactions can hamper efforts to map genetic variation to virus phenotypes , such as coreceptor usage , neutralization sensitivity , or drug resistance ., For example , a substitution at position 306 in HIV-1 gp120 ( relative to the HXB2 reference sequence ) is necessary , but not sufficient , to induce a shift in coreceptor usage in HIV-1; full expression of this phenotype requires additional substitutions at positions 320 or 324 15 , 16 ., Second , the identification of interacting residues may be applied toward defining a minimal set of HIV-1 protein sequences to be incorporated into a broadly reactive vaccine 17 , 18 ., Consequently , a substantial literature has developed around the goal of defining an accurate map of interactions in the HIV-1 envelope 19–22 ., The majority of these studies have focused on detecting interactions within the third variable domain of the external envelope glycoprotein gp120 ., The third variable domain ( V3 ) of the HIV-1 envelope typically spans 33 to 35 residues that are bounded by two invariant cysteines that form a disulfide bond to create a loop ., The V3 loop is characterized by extensive sequence variation , and is a principal determinant of important HIV-1 phenotypes such as coreceptor usage and cell tropism 23–25 ., Neutralizing antibodies elicited by the HIV-1 envelope are often directed against epitopes in the V3 loop 1 , 3 , and exposure to synthetic V3 peptides is sufficient to raise strain-specific neutralizing antibodies against lab-adapted strains of HIV-1 2 ., On the other hand , broadly reactive and potent neutralizing antibodies tend to recognize conformational rather than linear epitopes on V3 26 ., Because of its functional and antigenic importance , the three-dimensional structure of V3 has been studied extensively 27 , 28 , revealing a flexible , solvent-accessible loop that protrudes outward from the gp120 core toward the host cell ., To date , comparative studies of HIV-1 env V3 have looked for evidence of residue–residue interactions by measuring covariation among positions in a sample of nucleotide ( i . e . , codon ) or protein sequences 19–21 , 29 ., Sequence covariation is most frequently assayed by the application of one or more pairwise association test statistics , e . g . , mutual information 19 ., The resulting set of statistically significant pairwise associations is conventionally adjusted for multiple comparisons , either using the conservative Bonferroni correction 20 or the Benjamini-Hochberg false discovery rate ( FDR ) method 21 , 30 ., This procedure is straightforward and yields a set of putative interactions , but implicitly requires a number of unreasonable assumptions ., First , by treating each sequence as an independent observation in a random sample , the procedure ignores their evolutionary history ., However , it is well known that shared ancestry will produce spurious correlations between jointly inherited character states 31 , 32 ., This phenomenon has more recently been found to substantially alter the results of a landmark study into genetic associations of HIV-1 escape from cytotoxic T lymphocytes in a human population 33 , 34 ., Secondly , the pairwise associations that are selected by the test statistic have never been evaluated in the context of any other residue ., For example , an interaction between two residues may be dependent on the residue at a third position in the V3 loop ., Many of the test statistics employed by previous studies are inherently unable to model such higher-order associations , requiring that we assume such interactions do not exist ., Because each association is evaluated in complete isolation , we are left with a “laundry list” of pairwise associations with no apparent procedure for compiling these into a meaningful overall picture of interactions in the V3 loop ., In this study , we propose a new method for detecting interactions from an arbitrary sample of genetic sequences that relaxes both of these assumptions ., We apply our method to analyzing residue–residue interactions in the V3 loop of HIV-1 gp120 , which has emerged as a model system for the implementation of association test statistics or classification algorithms 16 , 19 , 20 , 29 , 35 ., Instead of quantifying covariation in the observed set of sequences D , we will take their evolutionary histories as our data 36–38 ., Because these data are almost always unobservable , they must be inferred from D by assuming that the sequences have evolved according to some stochastic model M . We will also assume that the phylogenetic tree T , which defines the common ancestry of extant sequences , is known ., Using maximum likelihood , we can infer the ancestral sequences D′ at each branching point , or node , of the tree T as a function of D and M 39 , 40 ., Any difference between sequences in D ∪ D′ occupying adjacent nodes of T implies that one or more substitution events occurred at that site in the intervening branch 41 ., The end result is a sample of evolutionary events encoded as a matrix D′′ , in which each unit of observation ( row ) corresponds to a branch in the phylogenetic tree onto which substitutions are mapped ., Each column of D′′ corresponds to a unique codon position , containing a “1” for every branch in which one or more substitutions occur at that position , and “0” otherwise ., In sum , D′′ is a phylogenetically independent sample of nonsynonymous substitution events that is derived by augmenting the observed data D with an evolutionary model and a tree ., To address the second assumption , we will analyze the phylogenetically augmented data D′′ as a Bayesian network ., A Bayesian network , B , is a graph that encodes a set of conditional independence assumptions on the joint probability distribution of random variables 42 ., A graph is a visual depiction of relationships between unique objects that typically assumes the form of points ( nodes ) connected by line segments or arrows ( edges ) ., In this case , each node represents a random variable whose outcome may depend on other variables ., For example , a directed edge originating from node A and terminating at node B ( A → B ) represents the probabilistic assumption P ( A ∩ B ) = P ( A ) P ( B | A ) ≠ P ( A ) P ( B ) , i . e . , that B is conditionally dependent on A . The set of all edges in the graph corresponds to the “structure” of the Bayesian network , BS ., Bayesian networks hold a considerable advantage over pairwise association tests ., First of all , pairwise association tests are unable to distinguish between direct and indirect associations between variables ., For instance , consider the case in which two nodes are each dependent on a third ( B ← A → C ) ., Association test statistics are susceptible to attributing significant associations to all three pairs ., However , B is conditionally independent of C , given A; i . e . , P ( B ∩ C | A ) = P ( B | A ) P ( C | A ) ., Because conditional independence is explicitly encoded by a Bayesian network , we can obtain a more parsimonious and informative representation of biological causation 43 ., Secondly , Bayesian networks provide an efficient representation of the joint distribution in an accessible graphical format ., It is therefore not necessary to assemble an ad hoc summary network from a list of significant pairwise associations ., We apply this “evolutionary-network” model to detect interactions among residues in the V3 loop of the HIV-1 envelope ., Using maximum likelihood , we infer a phylogenetically independent set of substitution events ., Interactions among residues are manifested as correlated substitutions within this inferred set , such that substitutions affecting a subset of residues tend to be mapped to the same branch of the tree ., Because the phylogenetic inference of substitution events is susceptible to some uncertainty , we carry out a parametric bootstrap procedure to quantify the sensitivity of the results from a maximum-likelihood reconstruction ., We also apply our method to several control cases , including the better-characterized compensatory interactions in HIV-1 protease , to validate our results for the V3 loop ., Our analysis reveals a large number of interactions among residues that fall into stratified clusters along the length of the V3 loop ., A total of 1 , 154 full-length sequences of HIV-1 env were obtained from the Los Alamos National Laboratory ( LANL ) HIV sequence database ( http://www . hiv . lanl . gov ) , excluding recombinant sequences and limited to one sequence per patient ., The nucleotide alignment was adjusted using Se-Al ( Andrew Rambaut , http://tree . bio . ed . ac . uk/software/seal ) ., According to the LANL subtype annotation , this alignment comprised 500 ( 43 . 3% ) subtype C , 431 ( 37 . 3% ) subtype B , 109 ( 9 . 4% ) subtype A , 65 ( 5 . 6% ) subtype D , 25 ( 2 . 2% ) subtype G , 17 ( 1 . 5% ) subtype F , three ( 0 . 3% ) subtype H , two ( 0 . 2% ) subtype K , and two ( 0 . 2% ) subtype J sequences ., Using the “11/25 rule” 16 , 44 , we predicted that 131 sequences encoded V3 loops binding the CXCR4 coreceptor; this subset comprised predominantly subtypes B and D ( n = 105 ) ., We used the nucleotide alignment to reconstruct a phylogeny by neighbor-joining 45 using Tamura-Nei distances 46 with rate variation across sites ( parameterized by a gamma distribution with a shape parameter α = 0 . 5 ) , excluding the indel-rich variable domains of gp120 to avoid the confounding effects of uncertainty in alignment ., We removed nine sequences that were identical in the portion of the alignment applied toward reconstructing the phylogeny ., We fit a codon substitution model 47 combined with the general time-reversible nucleotide substitution model ( GTR ) 48 to the alignment and tree by maximum likelihood using HyPhy 49 ., The branch lengths of the nucleotide tree were constrained to be scaled by a constant factor , reducing considerably the number of parameters to be estimated; this has been previously demonstrated to be a robust approximation for fitting codon models 50 , 51 ., Maximum-likelihood reconstructions of ancestral sequences were extracted from the fitted model for each internal node of the tree ., We inferred that one or more nonsynonymous substitutions had occurred along a branch if the reconstructed codon states at the nodes at either end of the branch encoded different amino acids 41 , 51 ., Each branch of the tree was thereby annotated with a binary-state vector , according to the presence or absence of a nonsynonymous substitution in that branch ., This procedure yielded a matrix comprising 2 , 305 rows and 33 columns ., To quantify the uncertainty in the reconstruction of ancestral sequences , we generated parametric bootstraps in HyPhy by resampling ancestral sequences in proportion to their likelihood 41 , 51 given the parameter estimates of the evolutionary model ( i . e . , branch lengths and codon substitution rates ) , resulting in 100 replicate matrices that were analyzed alongside the original matrix using Bayesian networks , as described in the next section ., We analyzed the matrix of substitution events mapped to branches in the phylogeny ( D′′ ) as a Bayesian network comprising 33 discrete nodes , using an algorithm proposed by Friedman and Koller 52 that we implemented in HyPhy ., The problem of detecting interactions among codon positions is equivalent to “learning” the structure BS of a Bayesian network B from data D′′ 53 , in which the structure refers to a set of directed edges representing conditional dependencies between nodes ., According to Robinsons 54 recursive formula , there are approximately 2 . 67 × 10190 possible network structures on 33 nodes; this number clearly precludes an exhaustive search for an optimal structure ., Furthermore , more than one network structure may be supported equivocally by the data , especially when the number of observations is small relative to the number of nodes 52 ., Friedman and Koller proposed carrying out a Markov Chain Monte Carlo ( MCMC ) algorithm over the space of node orders rather than network structures ., A node order is a permutation of the nodes in a linear sequence such that a node can only become assigned as a “parent” of nodes that are positioned to its left ( i . e . , “precedes , ” ≺ ) ., For example , the node order A ≺ B defines a subset of network structures that excludes all structures containing the edge A ↮ B . The node order space is relatively more compact—for instance , there are approximately 8 . 6 × 1036 permutations of 33 nodes—and yields a smoother posterior probability surface with improved MCMC convergence properties 52 ., Following Friedman and Koller 52 , our implementation precomputed the posterior probability , or score , for every combination of states assigned to the parental nodes of the ith node , for all i ., These node scores were cached into memory and accessed by direct indexing to make subsequent calculations more efficient ., The posterior probability of a structure BS was calculated according to the K2 scoring metric 55 , which integrated over the conditional probabilities at each node ( i . e . , the network parameters , BP ) that were distributed according to an uninformative Dirichlet prior , i . e . , a uniform distribution over the interval 0 , 1 for binary-valued nodes ., The K2 metric tends to favor more parsimonious , and hence more interpretable , network structures than alternative scoring metrics such as the Bayesian Dirichlet metric or BDe 56 ., ( The BDe metric yielded nearly identical results to the K2 metric when the “imaginary” prior sample size required by the BDe was set to near zero . ), Our use of cached node scores exploited the fact that the posterior probability of a network structure decomposes into a product of node “families , ” which blocks the peripheral structure by accounting for the direct influence of the parents of each node ., We ran a single Markov chain initialized with a random permutation of the node ordering ., At each step , a proposal function swapped two random nodes in the current node order ., The proposed node order was accepted unconditionally if its posterior probability exceeded that of the current ordering , or conditionally in proportion to the ratio of posterior probabilities ( i . e . , Metropolis-Hastings sampling 57 ) ., We ran the Markov chain for 106 steps with a burn-in period of 105 steps , which we have found to be more than sufficient for convergence for networks of this size ., We ran a duplicate Markov chain and found that Gelman and Rubins convergence diagnostic 58 was indistinguishable from one , which was consistent with convergence ., Node order statistics ( e . g . , edge posterior probabilities ) were sampled at 104 steps of the chain at equal intervals ., We estimated the posterior probability for each of 528 possible edges as the proportion of structures in the sample in which the edge was present in either direction , weighted by P ( BS | D ) 52 ., A consensus network structure was assembled from all undirected edges with a marginal posterior probability exceeding the cutoff value 0 . 95 ., The same analysis was carried out for each parametric bootstrap sample , except the chain in each case was run for 105 steps with a burn-in period of 104 steps and 103 samples ., The profile of each chain was visually inspected to evaluate convergence ., The frequency of edges with a posterior probability exceeding 0 . 95 was summed across bootstrap samples to quantify the sensitivity of edges in the maximum-likelihood consensus network to uncertainty in the reconstruction of ancestral sequences ., We employed three validation procedures to evaluate the accuracy of our methods ., First , we invoked a paired binary-character model , originally developed to analyze the evolution of N-linked glycosylation site motifs 22 , in order to simulate the evolution of V3 sequences along the “observed” phylogeny with a known set of interactions ., This model specifies the substitution rates between the paired states {00 , 01 , 10 , 11} , disallowing simultaneous substitutions affecting both sites in a pair , i . e . , 00 <≠> 11 , 01 <≠> 10 ., We converted our alignment of V3 sequences into binary-character sequences based on the presence or absence of the alignment consensus amino acid at each position of the sequence ., For example , any V3 sequence that encoded a threonine at position 1 was converted into a binary string beginning with “1 , ” and “0” otherwise ., We constrained the substitution rate parameters of the paired binary-character model such that the expected frequency of “1”s in simulated alignments would equal the observed frequency of consensus residues in our V3 alignment ., A “pairwise coevolution” parameter , ɛ , determined the factor by which a “1” at one site accelerated the rate of “0” → “1” substitutions at the second site , or conversely the odds that a “0” at one site became replaced by a “1” with respect to the presence or absence of a “1” at the second site of a pair ., In other words , if ɛ = 1 , then the paired model was effectively equivalent to a model in which every site evolved independently ., We simulated the evolution of binary sequences comprising 17 coevolving pairs of sites ( i . e . , {1 , 2} , {3 , 4} , . . . , {33 , 34} ) along the original neighbor-joining tree ., Two hundred replicate alignments were generated in this fashion for a range of ɛ parameter values ., Each alignment was analyzed by the evolutionary-network method outlined above , i . e . , by fitting a model of independently evolving binary characters by maximum likelihood , assigning substitution events to branches in the tree , and analyzing the distribution of substitutions as a Bayesian network using an MCMC analysis ., We recorded the frequency that edges in the network with marginal posterior probabilities exceeding 0 . 95 recovered our predefined paired interactions ( true positives ) or spurious ones ( false positives ) ., These results were contrasted with the rates of true and false positives obtained from using the Fisher exact test on the simulated binary sequence alignments ( i . e . , without correcting for the phylogeny ) ., We chose the Fisher exact test as a representative example of pairwise association test statistics ., Second , we simulated the evolution of nucleotide sequences along the tree according to a more realistic codon substitution model whose parameters were estimated from the original alignment of V3 sequences ., We randomly generated 100 replicate alignments with the same dimensions and characteristics ( e . g . , expected codon frequencies ) as our observed V3 alignment by this method ., Because the codon substitution model assumes that an alignment is a set of independently evolving codon sites 47 , any significant interactions between sites were false positives caused by founder effects in the phylogeny ., We evaluated the false-positive rate for our method against the rate for a more conventional pairwise association test , in which we applied the Fisher exact test to every pairwise combination of codon positions in the simulated V3 loop sequences , enumerating consensus and nonconsensus residues to generate a 2 × 2 contingency table ., Third , we applied the evolutionary-network model to a set of HIV-1 protease sequences , in which compensatory interactions are substantially better characterized empirically or structurally than the V3 loop , particularly in the context of drug resistance 11 , 12 , 59–64 ., We obtained an alignment of 2 , 641 HIV-1 subtype B sequences from the Stanford HIV resistance database ( http://hivdb . stanford . edu ) 65 representing patients on active drug regimens that included at least one protease inhibitor ., This alignment was analyzed using the evolutionary-network methods , and the results were contrasted with known interactions from the empirical and structural literature ., We mapped interactions from the consensus Bayesian network to a three-dimensional structure of the V3 loop of HIV-1 gp120 complexed to the CD4 receptor and X5 antibody ( Research Collaboratory for Structural Bioinformatics Protein Data Bank RCSB PDB ) , using the visualization software Chimera ( University of California San Francisco , Computer Graphics Lab 66 ) ., The maximum-likelihood reconstruction of ancestral sequences along the tree resulted in approximately 1 . 87 nonsynonymous substitutions per branch ., The mean number of inferred nonsynonymous substitutions was significantly divergent between internal ( 0 . 48 substitutions per branch ) and terminal ( 3 . 28 ) branches of the tree ( Wilcoxon rank sum test , W = 152041 , p ≪ 0 . 001 ) ; only 95 out of 1 , 142 ( 8 . 3% ) internal branches had more than one substitution mapped ., We found substantial variation in the total number of nonsynonymous substitutions among codon positions in V3 ( coefficient of variation , C . V . = 1 . 16 ) ., The largest number of inferred substitutions occurred at residue 24 , whereas residues 2 , 16 , 27 , and 32 were highly conserved ( numbered according to their position within the interval bounded by cysteines in the consensus sequence , as in 19 ) ., This distribution was consistent with the pattern of diversifying selection across sites ( unpublished data ) , implying that the differences were not simply due to variation among codon positions in mutation rate or the expected number of nonsynonymous sites ., Adjusting the inferred number of nonsynonymous substitutions for the expected number of nonsynonymous sites at each codon position , and normalizing by the analogous quantity for synonymous substitutions ( i . e . , dN − dS ) , indicated that residues 13 and 29 were even more strongly conserved than implied by the uncorrected frequency of nonsynonymous substitutions ., The detection of interactions among codon positions in the number of nonsynonymous substitutions did not require any such correction , however ., We validated the accuracy of our evolutionary-network model using three different controls ., First , we simulated the evolution of HIV-1 V3-like sequences along the original phylogeny as vectors of binary characters switching between consensus and nonconsensus residues ., Each consecutive pair of residues was constrained to coevolve according to an adjustable parameter ɛ , where ɛ = 1 corresponded to independently evolving sites ., We contrasted the performance of a binary-state analog of the evolutionary-network model , reconstructing substitution events by maximum likelihood , against the results from applying the Fisher exact test to the extant binary sequences ( Figure 1A ) ., After correcting for multiple comparisons using the Benjamini-Hochberg correction 30 , we found that the Fisher exact test resulted in a very high number of false positives increasing with the pairwise interaction parameter ɛ ( with means ranging from 42 to 100 false positives out of 544 negative instances over the range of ɛ from 1 to 3 ) ., Although the Fisher exact test appeared to recover more true positives on average than our evolutionary-network model for a given value of ɛ , our model sustained a substantially lower rate of false positives ( averaging five out of 544 negative instances ) ., As a result , our model converged to the most desirable outcome ( 100% true-positive rate and 0% false-positive rate ) with increasing values of ɛ , whereas the Fisher exact test diverged closer to the line of no discrimination ( i . e . , random guess; Figure 1A ) ., Similar results were obtained using a more conservative Bonferroni correction for the Fisher exact tests ., Second , we simulated the evolution of HIV-1 V3 sequences along the phylogeny using a more realistic codon-based substitution model ., Because this model assumes that each codon site evolves independently , the number of significant associations from each replicate simulation provided an estimate of the false-positive rate ., Using the Fisher exact test on the pairwise combination of amino acids in simulated sequences , we found false-positive associations between 160 . 8 out of 528 pairs on average ( 10% and 90% quantiles = 115 . 4 and 217 . 3 , respectively; Figure 1B ) ., This number of significant pairwise associations corresponded to an expected false-positive rate of about 30 . 4% ( 10% and 90% quantiles = 21 . 8% and 41 . 2% , respectively ) , predominantly due to the common ancestry of sequences ( i . e . , founder effect 34 ) ., In contrast , our evolutionary-network analysis yielded about four false positives on average ( 0 . 8%; Figure 1B ) , corresponding to a 40-fold improvement in specificity ., Third , we applied our evolutionary-network method to analyze HIV-1 subtype B protease sequences isolated from 2 , 461 patients undergoing drug regimens including at least one protease inhibitor 65 ., HIV-1 protease is better characterized structurally than the relatively flexible V3 loop of the envelope glycoprotein gp120 12 , 63 , 67 ., In addition , compensatory interactions between several sites in HIV-1 protease are extensively documented from clinical and experimental studies 11 , 12 , 59–64 , with greater consistency among studies than interactions in V3 ., We obtained a consensus network comprising 16 edges with marginal posterior probabilities exceeding a cutoff of 0 . 95 ( Figure S1 ) ., Two nodes representing the codon sites M46 and V82—prefixed with the alignment consensus residue and numbered according to their position in the HXB2 reference sequence—were highly connected by five ( L10 , V32 , T74 , V82 , L90 ) and four ( M46 , V48 , I54 , A71 ) edges , respectively ., Both M46 and V82 are well-known sites of mutations that interact with mutations at the other sites identified in this network in order to confer resistance to protease inhibitors and cross-resistance to multiple inhibitors 11 , 12 , 62 , 63 ., We also recovered a highly significant edge between the codon sites D30 and N88 , at which mutations have been jointly implicated in clinical data as conferring specific resistance to the inhibitor nelfinavir 68 ., In sum , we found strong concordance between our network and clinical and experimental studies of compensatory mutations in HIV-1 protease ., The prior probability of every potential edge was set to 0 . 5 ., Given our augmented dataset , the distribution of the posterior probabilities of edges was strongly U-shaped , with a distinct cluster of edges with probabilities exceeding 0 . 95 ( Figure 2 ) ., This outcome indicated that our phylogenetically augmented data matrix D′′ was sufficiently informative to distinguish network edges supported by the data from edges with little support ., The consensus network assembled from edges above our cutoff comprised five components , including one large network component connecting 11 nodes ( Figure 3 ) ., All putative interactions identified by the consensus network were “positive” ( odds ratio OR > 1 ) , such that a substitution at one residue significantly increased the probability of a substitution at a different residue in V3 ., We caution that because nonsynonymous substitutions on branches were relatively rare events , our analysis may have been subject to an intrinsic lack of power to detect negative interactions , i . e . , where the occurrence of one substitution excluded substitutions at other sites ., The parametric bootstrap support values for each edge in the consensus network are provided in Figure 3 as integer values ( between 0 and 100 ) adjacent to each edge ., These values quantified the sensitivity of each edge to uncertainty in the reconstruction of ancestral sequences ., Nine of the edges in the network had support values below a cutoff of 50; these edges were trimmed from the final consensus network ., The strongest association in the consensus network occurred between residues 5 and 7 ( OR = 155 . 6 ) , which jointly defined a conserved N-linked glycosylation site motif ( i . e . , NNTR ) ., Upon inspection , we found 28 phylogenetically independent events in which substitutions occurred along the branch , affecting both residues and disrupting the motif ., Because a substitution at either residue would have been sufficient to eliminate the N-linked glycosylation site motif , this association suggested the presence of additional constraints on V3 in the absence of glycosylation ., We also found evidence of an interaction between residues 5 and 30 ( OR = 53 . 4 ) ., Although these residues resided on the opposite strands of the V3 loop , they were roughly equidistant from the base ( Figure 4 ) , which may facilitate an interaction bridging the loop ., Two of the network components ( R12–F19 and I13–Q17 ) represented positive associations that were nested with respect to the secondary structure ( Figure 4 ) , which was consistent with stratification of the V3 loop ., Both putative interactions were located in the region identified as the “tip , ” which has been implicated in initiating gp41-mediated fusion with the host cell membrane , in addition to acting as the binding site for several monoclonal antibodies 28 ., A large component comprised associations among the nodes S10 , D24 , I25 , and I26 , which all mapped to residues in the stem region of the V3 loop 28 ., This component included a strongly supported association ( in 86 out of 100 parametric bootstrap samples ) between residues 10 and 24 , which has been detected in previous studies of covariation in V3 loop sequences 19 , 20 ., These residues have been implicated as strong determinants of coreceptor usage , i . e . , the 11/25 rule 44 , and act synergistically to alter the syncytium-induction phenotype 15 ., Associations between residue 24 and the adjacent residues 25 and 26 were blocked by residue 10 ( Figure 3 ) ., Indeed , when we repeated the analysis with a ban on edges between the nodes S10 and D24 , node D24 nonetheless failed to become incorporated into the same component as I25 and I26 ., Overall , putative interactions between adjacent residues were more the exception than the rule; the majority of the putative interactions tended to bridge opposite strands of the V3 loop ., We applied the 11/25 rule to classify 131 of the extant sequences as yielding CXCR4-binding virus , i . e . , having an “X4” phenotype ., Thirty-seven of the X4 sequences formed monophyletic groups , for which each common ancestor may have been interprete | Introduction, Materials and Methods, Results, Discussion, Supporting Information | The third variable loop ( V3 ) of the human immunodeficiency virus type 1 ( HIV-1 ) envelope is a principal determinant of antibody neutralization and progression to AIDS ., Although it is undoubtedly an important target for vaccine research , extensive genetic variation in V3 remains an obstacle to the development of an effective vaccine ., Comparative methods that exploit the abundance of sequence data can detect interactions between residues of rapidly evolving proteins such as the HIV-1 envelope , revealing biological constraints on their variability ., However , previous studies have relied implicitly on two biologically unrealistic assumptions: ( 1 ) that founder effects in the evolutionary history of the sequences can be ignored , and; ( 2 ) that statistical associations between residues occur exclusively in pairs ., We show that comparative methods that neglect the evolutionary history of extant sequences are susceptible to a high rate of false positives ( 20%–40% ) ., Therefore , we propose a new method to detect interactions that relaxes both of these assumptions ., First , we reconstruct the evolutionary history of extant sequences by maximum likelihood , shifting focus from extant sequence variation to the underlying substitution events ., Second , we analyze the joint distribution of substitution events among positions in the sequence as a Bayesian graphical model , in which each branch in the phylogeny is a unit of observation ., We perform extensive validation of our models using both simulations and a control case of known interactions in HIV-1 protease , and apply this method to detect interactions within V3 from a sample of 1 , 154 HIV-1 envelope sequences ., Our method greatly reduces the number of false positives due to founder effects , while capturing several higher-order interactions among V3 residues ., By mapping these interactions to a structural model of the V3 loop , we find that the loop is stratified into distinct evolutionary clusters ., We extend our model to detect interactions between the V3 and C4 domains of the HIV-1 envelope , and account for the uncertainty in mapping substitutions to the tree with a parametric bootstrap . | The third variable loop ( V3 ) of the human immunodeficiency virus type 1 ( HIV-1 ) envelope is a principal determinant of viral growth characteristics and an important target for the immune system ., Interactions between residues of V3 allow the virus to shift between combinations of residues to escape the immune system while retaining its structure and functions ., Comparative study of HIV-1 V3 sequences can detect such interactions by the covariation of sites in the sequence , which can then be used to inform vaccine development , but current methods for detecting such associations rely on biologically unrealistic assumptions ., We demonstrate that these assumptions cause an excessive number of spurious associations , and present a new approach that couples phylogenetic and Bayesian network models , and greatly reduces this number while retaining the ability to detect real associations ., Our analysis reveals that the V3 loop is stratified into discrete layers of interacting residues , suggesting a partition of functions along this viral structure with implications for vaccine development . | evolutionary biology, viruses, virology, computational biology | null |
journal.pcbi.1002126 | 2,011 | Towards a System Level Understanding of Non-Model Organisms Sampled from the Environment: A Network Biology Approach | Modelling the responses and compensatory adaptations of living organisms to a changing environment is extremely important both in terms of scientific understanding and for its potential impact on global health ., Although computational modelling of ecological systems has been utilised in ecotoxicology , the application of systems biology approaches to non-model organisms in general presents formidable difficulties , partly due to limited sequence information for environmentally relevant sentinel species ., Moreover , the number of samples and the depth of information available are often limited and there may be a lack of truly relevant physiological endpoints ., Thus , omics have proven effective in finding responses of aquatic organisms to model toxicants in laboratory-based experiments 1 but the environment poses a greater challenge as anthropogenic contaminants are present as complex mixtures and responses will additionally be dependent upon natural life history traits and other environmental factors ., Relatively few omics studies have focussed upon the ecotoxicology of environmentally sampled fish 2–7 ., Although we have previously shown 8 , 9 that expression of stress response genes could be used to distinguish fish from environmental sampling sites with different underlying contaminant burdens , this gave little insight to the health outcomes of these molecular differences ., In this context , identifying molecular mechanisms of compensatory and toxic responses from observational data ( reverse engineering ) , an approach that has been so successful in clinical studies and in laboratory model organisms , is highly challenging in field studies ., We addressed this challenge by developing a novel network inference strategy based on the integration of multi-level measurements of populations of fish exposed to a diverse spectrum of environmental pollutants ., This provides a useful model for a network biology approach generally applicable to non-model species and represents a breakthrough in the way we study the mechanisms whereby organisms respond to chemical exposure in the environment ., We directed our efforts towards modelling molecular networks representative of populations of the flatfish European flounder ( Platichthys flesus ) sampled from marine environments of North Western Europe , including locations significantly impacted by anthropogenic chemical contaminants ., The study integrated measurements representing a broad spectrum of samples characterized using transcriptomics , metabolomics , conventional biomarkers and analysis of chemicals in sediments from the sampling sites ., Previous studies have shown both anthropogenic contamination and higher prevalence of pre-neoplastic and neoplastic lesions in flounder from the Elbe estuary 10 and from the Mersey and Tyne 11 , together with elevated levels of hepatic DNA adducts at these sites 12 ., Data integration was achieved by implementing a systems biology framework for network reconstruction , starting from cross-species mapping of sequence information to the integration of multi-level datasets within a framework for network inference 13 and culminating in the identification of network modules predictive of physiological responses to chemical exposure , valuable for marine monitoring 14 ., The networks we identified demonstrate a remarkable parallel between human liver carcinogenesis and environmental effects on fish liver as well as revealing potentially novel adaptation mechanisms ., The broader application of network biology approaches to other non-model species sampled from the environment is therefore likely to profoundly change our understanding of how living systems are likely to adapt to complex environments ., An important assumption in many eco-toxicology studies is that the molecular states of organisms reflect their biological responses to complex chemical mixtures present within that environment ., Indirect evidence suggests that this hypothesis may be correct ., For example , consistent with previous studies 9 , we have identified genes and metabolites that were differentially expressed between environmental sites ( the results obtained are shown in detail in Table 1 and Text S1 , Tables S1 and S2 ) ., Many of these were either known to be associated with stress responses or were previously shown to respond to anthropogenic chemical contaminants in fish ., Although these results were encouraging they did not provide a direct link between molecular status and response to specific chemicals ., Since sediment chemistry data was available , we assessed whether chemical contaminant profiles could be inferred from gene expression data and whether these would at least partially match the known sediment composition ., Our analysis was performed by linking genes differentially expressed between each sampling site and the reference site , with chemical-gene relationships within the Comparative Toxicology Database ( CTD ) 15 ., The Alde estuary was chosen as the reference site due to its low concentrations of major anthropogenic chemical contaminants ( Table 1 ) , both in sediment and in flounder livers ., These significant associations may be regarded as predictive of the most important classes of chemicals exerting their biological effects upon flounder gene expression amongst the highly complex chemical mixtures within the sediments at these sites ., Results were consistent with the initial hypothesis , as where contaminants were highlighted both by chemistry and CTD analysis; sediment concentrations all exceeded the lower OSPAR ecotoxicological assessment criteria , except for PAHs at the Morecambe site ., At Brunsbuttel , elevated chromium and polychlorinated biphenyls ( PCBs ) ; at Cuxhaven , chromium , nickel , lead , zinc , polycyclic aromatic hydrocarbons ( PAHs ) and PCBs; at Helgoland nickel , zinc , manganese and PCBs , at Mersey PCBs; at Morecambe Bay arsenic , nickel and PAHs; and at Tyne arsenic and PCBs were all predicted by the CTD analyses and confirmed by chemistry data ( Table 2 and Table S3 ) ., PAHs were predicted at Morecambe and Cuxhaven , with the AhR-inducer beta-naphthoflavone predicted at Brunsbuttel , Helgoland and Tyne , consistent with our finding of CYP1A transcriptional induction at all sites in comparison with the Alde ., Additionally , Ingenuity Pathway Analysis ( IPA ) of all genes identified as significantly differentially expressed between sites showed significant associations with a number of toxicologically important processes and outcomes ( Table 3 ) ., As there were clear relationships between geographical location , chemical exposure and molecular profiles of flounder livers , we proceeded to reconstruct a network model representing the relationships between transcriptomic and metabolomic data , morphological measurements , protein biomarkers and microsatellite markers ( Figure 1 ) ., This network was constructed from all data , not limited to molecules that differed between sites ., Inspection of the resulting network ( Figure 2 ) showed that transcriptional and metabolic networks separated into two different areas of the network layout ., Interestingly , modules whose hubs were fish morphometric measurements occurred exactly at the interface between these two areas and these modules contained metabolite ( 46% ) and transcript ( 50% ) measurements as well as fish morphometric measurements ( 4% ) ., Different areas of the inferred network ( Figure 2 ) were characterised with different functional profiles ., The modules close to the interface with metabolism ( A ) showed enrichment ( FDR<0 . 05 ) for the annotation terms mitochondrion ( GO:0005739 ) , oxidoreductase activity ( GO:0016491 ) , endoplasmic reticulum ( GO:0005783 ) , protein folding ( GO:0006457 ) and antioxidant activity ( GO:0016209 ) , and the two KEGG pathways hsa03050:proteasome and hsa00480:glutathione metabolism ., The second sub-network ( B ) was enriched for immune response ( GO:0006955 ) and response to stress ( GO:0006950 ) ., The third sub-network ( C ) was enriched for proteolysis ( GO:0006508 ) and digestion ( GO:0007586 ) ., Each individual network module was tested for its ability to predict geographic sampling sites ( Figure 3 ) , the presence of parasites ( Figure 4A ) and the presence of any of the liver histo-pathological abnormalities shown in Table 1H ( Figure S2 ) ., Modules that were predictive of environmental sampling site were concentrated in two sub-networks ., The larger ( Figure 3 , area A ) was centred on the interface between metabolic and transcriptional networks and consequently included 14 modules consisting of morphometric indices as well as metabolites and transcripts ( 1% morphometric indices , 36% metabolite bins , 63% transcripts ) ., Modules that were predictive of parasitic copepod infection by Acanthochondria sp ., and Lepeophtheirus sp ., were similarly distributed in the network , but with additional modules localized in the sub-network B that were enriched in annotation to the immune response ., Modules that were predictive of infection by Anisakid nematodes ( Figure 4A ) displayed a different profile , being more concentrated within group B . Hierarchical clustering of the profile of modules that were predictive of parasite infections showed that responses to copepod infections by Lepeophtheirus and Acanthochondria clustered together and were distinguished from responses to infection by the Anisakid nematodes ., We have previously shown that there is a strong link between laboratory exposure to individual chemicals and flounder hepatic gene expression 8 ., It was therefore reasonable to hypothesize that genes differentially expressed in laboratory exposures may map onto modules predictive of sampling location ., Fishers Exact Test was therefore used to identify modules where genes differentially regulated as a result of single chemical laboratory exposures were over-represented ( these were determined by ANOVA FDR<0 . 05 over 16 day time-courses post-intraperitoneal injection ) ., Responses to lindane are shown as an example in Figure 4B ., This highlighted the temporal change in responses to toxicants , with the majority of overlapping modules occurring in both sub-networks A and B at early timepoints , followed by a shift towards sub-networks B and C at later timepoints ., We have previously shown 8 that this temporal change is associated with an early induction of transcripts for chaperones , phase I and II metabolic enzymes , oxidative stress and protein synthesis that diminishes by the later timepoints and is replaced by induction of protein degradation , immune-function and inflammation-related transcripts ., The results for all treatments are illustrated graphically in Figure S2 E to L . All treatments showed overlap with modules in group A , at the metabolite/transcript interface , and this was clearest for cadmium , that only affected this area , apart from one module in group C . All other treatments showed overlap between responsive genes and group B modules to varying extents and all except estradiol and cadmium overlapped with at least two modules within group C . These results are supported by our previous study 9 in which we found that employing transcripts altered during laboratory exposures to a range of individual toxicants improved predictivity of environmental sampling sites ., Having defined network modules predictive of geographical location , Ingenuity Pathway Analysis was used to elucidate the detailed structure of molecular pathways and their potential association with specific signatures of liver pathology ., We performed these analyses under the hypothesis that the underlying response to chemical exposure would be consistent with what is known of human liver molecular pathophysiology ., It was therefore expected that significant associations between the modules defined by our analysis and networks stored in the Ingenuity database would be informative of the underlying molecular mechanisms ., We indeed observed a remarkable overlap between modules predictive of geographical location and modules containing genes whose transcriptional profile has been previously associated with liver fibrosis , cirrhosis and hepatocellular carcinoma in mammals ., Modules whose component genes related to hepatotoxicity are shown in Figure 5 ., The major group of site-predictive modules shows significant overlap with modules relating to liver cholestasis and hepatocellular carcinoma , whereas the secondary group overlaps with liver fibrosis ., The annotation gained from Ingenuity , with key regulators inferred from networks based on interaction information , was combined and clustered in the TMEV software package using 5 different algorithms ., These show ( Table 4 ) that genes and metabolites, a ) involved in bile acid synthesis , transport and amino acid metabolism, b ) predictive of parasite infection, c ) linked to hepatocellular carcinoma , reproductive disorders and liver cirrhosis, d ) responding to oxidative stressors tert-butylhydroperoxide ( tBHP ) and cadmium , the hormone estradiol and rodent peroxisome proliferator perfluoro-octanoic acid ( PFOA ) are closely linked to differences between environmental sites ., Additional relationships with inflammation , immune response , energy , fatty acids and nucleic acid metabolism , response to other toxicants and regulation by insulin , huntingtin , MYC and hepatocyte nuclear factor HNF4A were also highlighted ., Functional analysis of the modules that were both site-predictive and associated with hepatocellular carcinoma showed significant overlap with mitochondrion , proteasome , tricarboxylic acid cycle , melanosome , protein dimerization activity , membrane-enclosed lumen , glutathione metabolism , coenzyme binding , microsome , translation , protein transport and carbohydrate catabolism ( enrichment score >2 , FDR<0 . 05 ) ., The models we have developed are a high level representation of the molecular networks underlying response to environmental exposure ., In order to generate specific hypotheses on the molecular pathways modulated during compensatory adaptation and toxicity further in-depth analyses of the specific interactions between genes and metabolites were performed ., In this context , we combined the genes and metabolites represented in each group of predictive modules ( Groups A and B in Figure 3 ) and input these to IPA software ., The most statistically significant networks derived from each group are shown in Figure 6 and Figure S3 , coloured by expression represented as a ratio between a highly polluted site and the reference site ( Brunsbuttel versus Alde ) ., The component genes and metabolites were clustered and the resulting expression profiles are shown in Figure S1 ., The Ingenuity networks are further described in Text S1 and are discussed below ., The chemical exposures predicted from CTD interactions were partly confirmed by chemical data ( Table, 2 ) despite the complexity of the environment , potentially including mixture effects , bioaccumulation and non-chemical stressors ., Additional stressors were indicated that had not been chemically measured ., Taking the Brunsbuttel site as an example , ethinyl-estradiol was a predicted contaminant and serum VTG protein , a canonical marker of endocrine disruption , was induced relative to the Alde ( Table 1 ) ., Perfluorooctane sulfonic acid 16 and other persistent organic pollutants including PCBs , dieldrin and endosulfan 17 have been detected at elevated concentrations in the Elbe estuary and floodplain , and were all identified by our approach ., Additional chemicals highlighted included systhane and vinclozolin fungicides , the halogenated aromatic hydrocarbon pesticide lindane , chlorine and tetradecanoylphorboyl acetate ( TPA ) ., It is uncertain whether these compounds are in fact present at this site as , for example , the presence of TPA appears unlikely ., However TPA is a well-known tumour promoter 18 , so detection of its associated gene expression changes might be viewed as a biomarker of effect , not necessarily of a specific exposure ., At a number of sites flavonoids and flavonols , such as epicatechin gallate , were predicted , potentially indicating plant-derived exposures not of anthropogenic origin ., At Morecambe Bay and the Tyne the prediction of paraquat perhaps reflected an oxidative stress response rather than the presence of this particular compound ., These results support the use of a knowledge-based approach to infer chemical exposure profiles from molecular responses and validate the underlying assumptions in the study ., Predictions from interrogation of the CTD database ( Table 2; Table S3 ) differed between sites suggesting that the approach can be sufficiently sensitive to specific differences in the exposure profiles ., However , we do not propose that these associations necessarily indicate the presence of each specific contaminant at each site , for example ‘tobacco smoke pollution’ in the Mersey , we instead hypothesise that these represent the effects of related stressors , for example , AhR inducers at the Mersey site ., The development of a modular network , representing the integration between molecular and physiological readouts , provided us with an interpretive framework to analyse the complex molecular signatures linked to exposure ., One of the most interesting findings is that the modules that predict environmental exposure with greatest accuracy represent the interface between metabolite and transcriptional networks and link to higher level indicators of fish health , such as condition factor and hepatosomatic index ( Figure 2 ) ., Consistent with this observation , network modules at the interface between metabolite and transcriptional networks were also differentially regulated in response to single chemical laboratory exposures ., It should be borne in mind that the environmentally sampled fish have been chronically exposed to pollutants , and that chronic exposure can result in different responses than acute exposure 19 , 20 ., In addition bioavailability , mixture effects , metabolism and bioaccumulation affect compound-specific responses within the livers of these fish ., This is illustrated by the modules containing genes that responded to 16-day treatments of flounder with individual toxicants ( Figure 4 B , Figure S2 ) ., While all toxicants induced changes in the metabolic-interface genes , they also affected the secondary area of the network that related more to acute stress and immune response ( Figure 2 , area B ) , in contrast to the differences between environmental sites , where only one module ( 40 ) in this area was affected ., The characterisation of transcripts and metabolites that differed between sites was undertaken to provide insights into the molecular mechanisms that they describe , and to inform on the potential health outcomes for the fish ., Canonical pathways that contributed to these differences included those relevant to metabolism of toxicants; AhR signalling , metabolism of xenobiotics by cytochromes P450 , the NRF2-mediated oxidative stress response , glutathione metabolism and bile acid bioysnthesis ( Table 3 ) ., Together these describe phase I and II metabolism of xenobiotics , such as aromatic hydrocarbons , and their excretion via the bile ., Additional endobiotic metabolic pathways were affected ., Changes in glycolysis , pyruvate metabolism , the citric acid cycle and oxidative phosphorylation implied disturbances to the energy pathways of the liver that could reflect the energetic requirements of xenobiotic metabolism and lead to further metabolic disruption ., Changes in amino acid synthesis and proteasomal protein degradation also indicated reorganisation of metabolism ., This change in metabolic state and gene expression could be viewed as a successful compensatory response to toxicants and thus of little concern for the health of individual fish and these fish populations ., Further examination of the annotation of transcripts and metabolites differing between sites implied that this hypothesis was false ., As illustrated in Figure 5 , and shown in Tables 3 and 4 , there is a remarkable overlap between site-predictive modules and modules associated with hepatocellular carcinoma ( HCC ) ., Additionally , liver cholestasis -annotated modules overlapped with HCC and site predictive modules and this area of the network was highly associated with bile acid biosynthesis ., Apart from this metabolic interface group only one other module ( module 40 ) was predictive of site ., This was also associated with hepatocellular carcinoma , and additionally with liver fibrosis , indicative of chronic liver damage , and occurred in an area of the network associated with inflammation ., Therefore flounders inhabiting differentially contaminated sites show transcript and metabolite changes that have been associated with liver carcinogenesis in mammals ., A question remains as to whether this simply represents the detection of HCC in the liver samples , as histopathology data were unavailable for the fish sampled off Germany ., By comparison with studies of tumours from the closely related flatfish dab ( Limanda limanda ) this does not appear to be the case ., In dab tumours the metabolites choline , phospocholine and glycine were reduced in concentration and lactate increased , an indication of the switch to anaerobic metabolism in the bulk tumours 21 ., In , for example , the Brunsbuttel samples compared with non-tumour bearing Alde fish , choline , phosphocholine and glycine increased , and lactate decreased ., Additionally , transcripts for ribosomal proteins showed co-ordinated induction in bulk tumours from dab , indicative of proliferation 22 , but no such induction was apparent from the present samples ., The changes in gene expression and metabolites detected in this study do not recapitulate those found in bulk tumours , and may be viewed as indicating either an earlier stage of tumourigenesis or a permissive micro-environment in which hyperplastic tissue may form and lead to tumour formation ., Ingenuity networks , based on mammalian interaction data , permitted more detailed biological characterisation of the site-associated modules ., Complete pathways were not recapitulated by these analyses , as only a minority of the transcripts and metabolites from flounder liver were examined ., Nevertheless , the analyses highlighted important processes and inferred key regulators ., Here the most significant network derived from site-predictive modules is discussed in detail and additional networks are discussed in terms of their key inferred regulators ., The most striking finding from the Ingenuity analyses was the co-ordinated repression of proteasomal subunit genes at the Brunsbuttel site ( Figure 6A; Figure S3A1 ) ., This was not so marked at other sites ( Figure S1 ) , indeed at Morecambe Bay these genes were induced in comparison with the Alde fish ., Proteasome maturation protein ( POMP ) has been found to be a critical regulator of proteasomal activity 23 and has been shown to be repressed by the halogenated aromatic hydrocarbon 2 , 3 , 7 , 8-tetrachlorodibenzodioxin ( TCDD ) in an AhR-independent manner 24 ., Although TCDD concentration was not measured , the mean expression of proteasomal genes was inversely correlated ( r\u200a=\u200a−0 . 79 ) with fish liver PCB concentrations but did not correlate well with sediment PAH or PCB concentrations ., Tyne fish , for example , displayed relatively high proteasomal gene expression and had low liver PCB but high PAH concentrations ( Table 1 , Figure S1 ) ., Therefore the repression of proteasomal genes may represent a halogenated aromatic hydrocarbon-related response ( Figure 7 ) ., In trout Oncorhynchus mykiss , a proteasome inhibitor reduced PAH-dependent CYP1A induction 25 , in contrast to mammalian studies 26 ., This difference may contribute to the lower inducibility of CYP1A in flounder in comparison with many mammals ., Ingenuity analysis also predicted an interaction between the proteasome and NF kappa B , a key regulator of mammalian hepatocarcinogensis 27 ., The proteasome represses NK kappa B activation , and potentially disruption of proteasomal activity could have extensive additional effects on intracellular protein levels due to its role in the degradation of numerous proteins ., We found no significant changes in NF kappa B gene expression between sites , and the consequences of putative activation at the Brunsbuttel site , and repression at the Morecambe site , due to changes in the proteasome , are difficult to predict , as in the early stages of carcinogenesis NF kappa B can have a protective effect , whereas in later stages it can promote tumourigenesis 27 ., From the Ingenuity networks a number of key regulatory molecules were inferred ., These included insulin ( Figure 6B , Figure S3A2 , S3A4 , S3B1 ) , estrone , luteinizing hormone ( LH ) and follicle stimulating hormone ( FSH ) ( Figure S3A3 and S3A6 ) , platelet derived growth factor beta ( PDGFBB ) ( Figure 6B , Figure S3A2 , S3B1 ) , transforming growth factor beta ( TGF-beta ) ( Figure S3A8 ) , vascular endothelial growth factor ( VEGF ) ( Figure 6B , Figure S3A7 , S3B1 ) , tumour necrosis factor ( TNF ) ( Figure S3A5 ) , and angiotensinogen ( Figure 6B , Figure S3B1 ) ., Insulin , in fish as in mammals , is a key hormonal regulator of energy , glucose and lipid metabolism , all pathways that were identified as affected by sampling site ., By the Ingenuity networks it was linked to protein kinases , metabolites ( including glucose and lactate ) and the glucose transporter SLC2A4 ., The most obvious explanation for changes in insulin and related parameters would be differences in diet between fish from different sites ., Amino-acid levels are more important regulators of insulin in carnivorous fish such as the flounder than sugars 28 ., Dietary parameters would be expected to be highly variable depending upon recent feeding history of the fish , which was unknown for these individuals ., However , insulin can also be modulated by exposure to toxicants including organophosphates 29 that was suggested to lead to an increase in lipogenesis , in agreement with our observations of phospholipidosis in fish from polluted sites ( Table 1 ) ., Mild estrogenic endocrine disruption was suggested by VTG induction in Brunsbuttel fish , and networks shown in Figures S3A6 and S3A3 inferred that estrogen receptor alpha ( ESR1 ) , FSH and LH target genes were modulators of the different responses between sampling sites ., ESR1 and HNF alpha were linked in Figure S3A6 and both are involved in hepatic cholestasis , indeed EE2-induced hepatotoxicity has been linked to alterations in bile acid biosynthesis in mice 30 ., PDGFBB is the dimeric form of platelet derived growth factor beta ( PDGF-B ) ., Notably , PDGF-B over-expressing mice spontaneously developed liver fibrosis 31 , and PDGF-BB was inferred as part of the network deriving from the liver fibrosis-annotated module 40 in our analysis ., Additionally PDGF-B over-expressing mice developed hepatocellular carcinoma in response to phenobarbital and diethylnitrosamine treatment and induced TGF-beta and VEGF expression ., TGF-beta was inferred to be an important regulator in site-specific responses ( Figure S3A8 ) and is a well-known mediator of cancer initiation , progression and metastasis , via interaction with the inflammatory response 32 ., Furthermore , the pro-inflammatory cytokine TNF-alpha , an initiating signal for the innate immune response in fish as well as mammals 33 , was also identified by Ingenuity analysis ( Figure S3A5 ) ., Release of TNF alpha from Kupffer cells leads to hepatocyte cell death , regeneration and fibrosis that can lead to hepatocellular carcinoma 34 ., VEGF , best known as a stimulator of angiogenesis , was also highlighted in both the fibrosis-related and carcinoma-related sections of the network , and was linked with cell cycle , oncogenes and tumour suppressor genes ( CDKN1A , TP53 , MYC ) ., Angiogenesis is a key requirement for the transition from fibrosis to hepatocellular carcinoma 35 ., Angiotensinogen ( AGT ) is the precursor of angiotensin and was found to be repressed at all sites in comparison to the Alde reference site ( Table S1 ) ., Angiotensin is a signal for vasoconstriction in mammals and in fish its expression is related to osmoregulation 7 with repression in liver in response to higher salinity ., As the sampling sites differed in salinity , alteration of AGT transcription was not a surprising finding ., As shown in Figure 6B , AGT was a member of the fibrosis-related module 40 and was predicted to form part of a complex network with VEGF , PDGF and intracellular kinases ., Angiotensin has indeed been linked to stimulation of inflammatory liver fibrosis 36 , via fibroblast proliferation and production of inflammatory cytokines and growth factors , including TGF-beta ., Inhibition of the angiotensin system by antagonism of its receptor 37 or inhibition of angiotensin-converting enzyme 38 has been shown to reduce hepatic fibrosis ., VEGF , TGF beta , TNF alpha , PDGF and AGT are all intimately related to the progression of fibrosis to cirrhosis and hepatocellular carcinoma in mammals ., These molecules were all highlighted as important regulators of the differences between molecular profiles of flounder livers from different sampling sites using an unbiased approach combining network inference and predictive algorithms ., A combination of omics , multiple biomarkers and bioinformatics were used to identify and characterise hepatic molecular changes between fish sampled from several environmental sites ., Based on these data , parasite infection , fish morphology and genetics do contribute to the differences between sites , but do not explain the majority of changes seen ., For example , within-site tests showed that morphometric parameters and parasite infections could be significantly associated only with a small proportion ( <3% ) of the gene expression differences between sites ( Table S1 , Table S4 ) ., Taken as a whole with our previous studies 8 , 9 , we find that anthropogenic chemical contamination of the marine environment is a major factor in explaining the molecular differences between fish sampled from these sites ., The different methodologies employed displayed different strengths and weaknesses ., Histopathology was a good guide to broad levels of pollution effect , but provided little information upon the nature of the contaminant profile ., Protein biomarkers and enzyme activities were useful for categorising sites by major classes of toxicant , but gave little information on the potential health outcomes ., 1H NMR metabolomics showed low technical variability , and metabolite profiles alone were more predictive of sampling site than gene expression profiles alone , however the annotation of metabolites is not yet well advanced , limiting the functional information currently available ., Transcriptomics exhibited higher variability than metabolomics , but was more informative due to better annotation ., Overall the methodologies were highly complementary , allowing analyses that would be impossible if one were limited to a single technique ., The gene expression signatures associated with fish from each sampling site were used to predict the presence of chemical contaminants using the CTD gene expression-chemical interaction database ., Mixture effects , other environmental influences and the similarity of certain stressors , such as the metals , might be expected to confound this approach ., Additionally the incomplete nature of the flounder microarray and the CTD database and the limited numbers of samples for certain sites , which is a common issue in field studies , reduce the potential of this analysis ., Therefore we did not expect to predict all environmental contaminants by this method ., While this approach was useful with the current dataset , it may be expected to improve in future as both the CTD database and transcriptomic data become more c | Introduction, Results, Discussion, Methods | The acquisition and analysis of datasets including multi-level omics and physiology from non-model species , sampled from field populations , is a formidable challenge , which so far has prevented the application of systems biology approaches ., If successful , these could contribute enormously to improving our understanding of how populations of living organisms adapt to environmental stressors relating to , for example , pollution and climate ., Here we describe the first application of a network inference approach integrating transcriptional , metabolic and phenotypic information representative of wild populations of the European flounder fish , sampled at seven estuarine locations in northern Europe with different degrees and profiles of chemical contaminants ., We identified network modules , whose activity was predictive of environmental exposure and represented a link between molecular and morphometric indices ., These sub-networks represented both known and candidate novel adverse outcome pathways representative of several aspects of human liver pathophysiology such as liver hyperplasia , fibrosis , and hepatocellular carcinoma ., At the molecular level these pathways were linked to TNF alpha , TGF beta , PDGF , AGT and VEGF signalling ., More generally , this pioneering study has important implications as it can be applied to model molecular mechanisms of compensatory adaptation to a wide range of scenarios in wild populations . | Understanding how living organisms adapt to changes in their natural habitats is of paramount importance particularly in respect to environmental stressors , such as pollution or climate ., Computational models integrating the multi-level molecular responses with organism physiology are likely to be indispensable tools to address this challenge ., However , because of the difficulties in acquiring and integrating data from non-model species and because of the intrinsic complexity of field studies , such an approach has not yet been attempted ., Here we describe the first example of a global network reconstruction linking transcriptional and metabolic responses to physiology in the flatfish , European flounder , a species currently used to monitor coastal waters around Northern Europe ., The model we developed has revealed a remarkable similarity between network modules predictive of chemical exposure in the environment and pathways involved in relevant aspects of human pathophysiology ., Generally , the approach we have pioneered has important implications as it can be applied to model molecular mechanisms of compensatory adaptation to a wide range of scenarios in wild populations . | genome expression analysis, functional genomics, environmental sciences, marine monitoring, marine biology, toxicology, environmental protection, marine and aquatic sciences, regulatory networks, biology, microarrays, systems biology, predictive toxicology, biochemistry, ecology, earth sciences, genomics, computational biology, genetics and genomics | null |
journal.pbio.1001133 | 2,011 | The Glial Regenerative Response to Central Nervous System Injury Is Enabled by Pros-Notch and Pros-NFκB Feedback | The structure of organisms is robust ., Cells accommodate changes in their environment during development and throughout life by adjusting cell number and cell morphology to preserve overall organismal integrity ., In the central nervous system ( CNS ) , adjustments are carried out in interacting populations of neurons and glial cells , from development to learning , ultimately enabling function ., Injury and regeneration experiments and theoretical models have long been used to uncover the cellular and molecular mechanisms of how cells “sense” and maintain normal organ structure 1 ., The premise is that shared mechanisms may underlie normal structural homeostasis and plasticity , and cellular responses to injury ., Understanding such mechanisms is one of the frontiers in biology ., It will also lead to a greater understanding of regeneration and repair of relevance to the treatment of human injury and disease ., The fruitfly Drosophila is an ideal model organism for discovering gene networks , and has been successfully used to investigate cellular responses to CNS injury 2–6 ., Here , we use Drosophila to uncover a gene network that controls the glial regenerative response to injury and promotes robustness in the normal CNS ., Previous experiments had revealed two important findings about glial responses to injury in Drosophila ., Firstly , enwrapping glial cells become phagocytic upon injury clearing cellular debris ., This phagocytic function requires the corpse engulfment receptor Draper , which is specifically expressed in enwrapping glia , and whose function involves Simu , Src42A , and Shark 4 , 6–9 ., Secondly , stabbing injury in the adult head 5 and neuronal ablation in the embryonic Ventral Nerve Cord ( VNC ) 10 induce the proliferation of glial cells ( including the enwrapping Longitudinal Glia ) ., Similar findings had long before been observed in the cockroach 11–13: surgical lesioning and chemical ablation of enwrapping glial cells induced cell transformation leading to the phagocytosis of cellular debris , and most remarkably glial proliferation ., This restored glial numbers , enwrapment , and normal electrophysiological function ., Insect glia may be evolutionarily distant from mammalian glia , but it can be insightful to compare them ., Injury induces distinct responses in mammalian CNS glial cells ., Astrocytes normally maintain ionic homeostasis , provide nutrients , and participate in synapses ., Microglia are the immune cells of the brain , and are normally in a resting state ., Upon injury , microglia and astrocytes phagocytose degenerating axons and other cellular debris , and they can also form a glial scar that inhibits axonal regeneration 14 ., Ensheathing glia ( oligodendrocytes ) normally myelinate axons for saltatory conduction in vertebrates , maintain ionic homeostasis , and provide metabolic and trophic support to axons 15 , 16 ., Oligodendrocyte progenitor cells ( OPCs ) respond to injury by dividing , and resulting oligodendrocytes remyelinate 17–19 ., This latter response is regenerative , leading to spontaneous re-enwrapment of axons and partial functional recovery , for instance of locomotion 20–22 ., Conditions such as spinal cord injury , stroke , and multiple sclerosis induce the proliferation of OPCs resulting in spontaneous remyelination of CNS axons , and underlie the “remission” phases of multiple sclerosis 18 , 23–28 ., Thus , the regenerative response of ensheathing glia occurs across the animals , from insects 11 to fish 23 and humans 20 ., In cockroach , fruitflies , and vertebrates , ensheathing glia proliferate upon injury , and both in insects and mammals this response can lead to limited remyelination and some recovery of function ., This reveals that there is an endogenous tendency of the CNS to repair itself ., Its manifestation across species may reflect a common underlying gene network ., If understood , it could be harnessed to stimulate CNS repair ., Here , we search for a Glial Regenerative Response ( GRR ) gene network that can promote repair after injury and confer structural robustness in the normal animal ., The following factors are promising candidates to belong to this gene network ., The Drosophila TNF super-family member Eiger triggers the proliferation of adult brain glia upon injury in fruitflies 5 ., TNFα also triggers the proliferation of mammalian oligodendrocytes progenitors through its receptor TNFR2 upon injury 29 ., While in other contexts TNFR2 is thought to function by activating NFκB 30 ( which can promote the cell cycle ) , whether this is the case for CNS glial cells and whether this activates glial proliferation are unknown ., Notch maintains the undifferentiated and stem cell state in many contexts 31 ., In Drosophila , Notch maintains the mitotic potential of embryonic ensheathing glia in interaction with the Jagged1 homologue , Serrate , from axons 10 , 32 ., Similarly , in vertebrates , Notch1 maintains the oligodendrocyte progenitor state by interacting with its ligand Jagged1 present in axons 33 ., However , the functions of Notch in the glial regenerative response remain unsolved ., Notch1 is present in adult NG2+ OPCs , and it is upregulated upon injury and during regeneration , but conditional Notch1 knock-out in OPCs does not prevent the regenerative response 34 , 35 ., Notch1 can inhibit the differentiation of progenitors into myelinating oligodendrocytes , preventing repair 33 , 36 , but presence of Notch1 signaling in these cells does not prevent the regenerative response either 34 , 35 ., Finding out how to control Notch1 function , to enable glial proliferation , and to subsequently promote ensheathing glial differentiation is thus a key issue ., The transcription factor Prospero ( Pros ) interacts with Notch in ensheathing glia in Drosophila embryos 10 , but how Pros and Notch affect each others function is not understood ., Pros appears to have opposite functions in neuroblasts and in glia ., In neuroblasts Pros is a tumour suppressor , as mutations in pros result in over-proliferation 37–40 ., Instead in glia both Pros and Notch are necessary for proliferation , and mutations in pros do not result in glial hyperplasia 10 , 32 ., Unravelling the relationship between Notch and Pros may hold the key to understanding how glial proliferation and differentiation are regulated ., Here , we uncover the Glial Regenerative Response gene network in Drosophila ., For this , we establish a new CNS injury paradigm in the larval VNC ., We use the larva because it is more accessible than the adult while it has locomotion , senses , learning , and memory , enabling the investigation of repair in the context of a fully functional CNS ., We show that the ensheathing Interface Glia of the VNC respond to injury by phagocytosing and clearing cellular debris and by dividing ., We reveal a gene network that controls the balance of glial proliferation and differentiation , and it is comprised of two feedback loops: one involving Pros and Notch , and a second involving Pros and Dorsal/NFκB , connected via Eiger/TNF and Wgn/TNFR signaling ., By manipulating this gene network we could shift from exacerbating the damage to promoting repair of the damaged neuropile ., The uncovered gene network is a homeostatic mechanism for structural robustness and plasticity ., To test how Drosophila larval glia respond to injury , dissected ventral nerve cords ( VNCs ) were stabbed with a fine needle and cultured ( Figure 1A ) ., Stabbing was applied dorsally into the neuropile , which comprises the bundles of CNS axons and Interface Glia ( IG ) 41 ., To obtain an overview of the injury response , we performed a time-lapse analysis ., Axons were visualised with GFP driven by the protein trap line G9 , and glia with repoGAL4>DsRed ( Figure 1B , Video S1 and Video S2 ) ., The wound initially expanded , and vacuoles formed within the neuropile ., However , after 6 h of culture , glial processes invaded the wound , the axonal and glial wound began to shrink , and by 22 h the wound could heal considerably ., This suggested that there is a natural mechanism that can promote repair ., Here ( Figure 1C ) , we ( 1 ) characterised the glial responses to injury , ( 2 ) investigated the gene network controlling the regenerative potential of glia , and ( 3 ) tested whether altering the functions of this gene network could promote glial regeneration and axonal repair ., To characterise the glial responses to injury , glial membranes were visualised with repoGAL4>mCD8-GFP and Interface Glia ( IG ) with anti-Glutamine Synthetase 2 ( GS2 ) ( Figure 2A , B and Figure S1 ) ., This revealed the stabbing wound dorsally and an indentation ventrally ( Figure S1 , Video S3 and Video S4 ) ., Although stabbing damaged to some extent surface and cortex glia ( unpublished data ) , it affected most prominently the IG , causing GS2+ glial loss ( Figure 2A , B and Figure S1 ) and GS2+ glial debris at 6 h after injury ( Figure 2B ) ., We used the alrmGAL4 driver , which is restricted to the Pros+ IG ( Figure S2A ) , to visualize IG nuclei with HistoneYFP , and this showed that some IG were lost through apoptosis ( Figure 2C , intact VNCs had 0 cleaved-Caspase-3+ YFP+ IG n\u200a=\u200a10 VNCs versus stabbed VNCs with an average of 1 . 5 cleaved-Caspase-3+ YFP+ in 57% of the VNCs n\u200a=\u200a14 ) ., In remaining IG , injury provoked an increase in the size and complexity of cytoplasmic projections ( seen with the membrane reporter mCD8GFP , Figure 2D ) ., As observed in time-lapse , injury led to vacuolization of the neuropile , as holes formed within the axonal bundles ( Figure 1B and Figure 2E ) ., IG projections enveloped these vacuoles ( Figure 2E ) ., To further characterise these aspects , we used Transmission Electron Microscopy ( TEM ) ., In wild-type wandering larvae , the IG nuclei were located outside the neuropile and their cytoplasms enwrapped the entire neuropile ( Figure 3A ) ., IG processes projected into the neuropile , where they could enwrap smaller axonal bundles ( Figure 3B ) and individual axons ( Figure 3C ) ., As seen with confocal microscopy , TEM confirmed that injury caused glial loss with breakdown of neuropile enwrapment after 6 h ( Figure 3D ) , and vacuoles formed within the neuropile ( Figure 3D ) ., Some Interface Glia degenerated via necrosis as seen by swelling of mitochondria ( Figure 3E ) ., Remaining IG expanded their cytoplasmic projections both around and within the neuropile ( Figure 3F–J ) , which was never observed in intact specimens ., IG processes lined vacuoles ( Figure 3F , G ) , phagocytosed axonal fragments and other cellular debris , as revealed by phagosomes and multilamellar bodies within the glial processes ( Figure 3H–J ) ., IG processes frequently wrapped around isolated axons that could be degenerating ( Figure 3K–N ) ., These data show that upon injury IG phagocytose cellular debris , presumably clearing the lesion ., Altogether , these data demonstrate that larval IG enwrap CNS axons , and that they are damaged by , and respond to , injury ., Next , we investigate if stabbing induced IG proliferation ., Based on their location , the IG are classified into dorsal ( dIG ) , lateral ( lIG ) , and ventral ( vIG ) IG ( Figure 4A ) ., They are identified by co-localisation of the pan-glial marker Repo and the transcription factor Pros in nuclei , surrounded by the cytoplasmic marker Ebony ( Figure S2B , C ) ., IG do not normally divide 42 , but are arrested in G1 and have mitotic potential ( Text S1 and Figure S3 ) ., To examine cell proliferation , we used PCNA-GFP , a reporter with E2F binding sites that reveals GFP expression when cells go through S-phase ( Figure 4B , C , D ) 43 ., PCNA-GFP+ Pros+ Ebony+ IG were rarely seen in non-stabbed controls , but stabbing increased their frequency at the lesion site ( Figure 4B ) and throughout the neuropile ( Figure 4C ) ., We did not find any Ebony-negative IG with PCNA-GFP ( unpublished data ) , suggesting that Ebony+ Pros+ IG are the only IG that divide in response to injury ., Normally there is one Ebony+ vIG per hemisegment , but the number of Ebony+ vIG adjacent to the wound increased significantly in stabbed larvae ( Figure 4E , F ) ., Altogether these data show that stabbing causes a local increase in proliferation of Pros+ Ebony+ IG at the lesion site ., A BrdU pulse experiment ( a commonly used method to visualise proliferating cells ) also revealed an overall increase in the number of dividing IG upon stabbing , to 50% of Ebony+ IG being also BrdU+ ( p<0 . 05 ) , comprising local IG at the lesions site ( Figure S4 ) and at some distance along the VNC ., This suggested that neuronal damage may also affect glial cells at a distance from the original lesion , which is explained as axons extend along the whole length of the VNC ., Stabbing may also affect other glial classes than IG ., To take these facts into account , we purposely developed DeadEasy Glia software to automatically count in vivo all Repo+ glial cells ( Figure 4G and Figure S5 ) ., After 22 h culture , stabbed VNCs had more glial cells than non-stabbed controls ( Figure 4G ) ., This effect was abolished when a cell cycle inhibitor—constitutively active Retinoblastoma protein factor ( Rbf280/Rb ) —was expressed in glia ( Figure 4G ) ., This demonstrates that the increase in glial number upon stabbing was due to the induction of glial proliferation ., These data show that stabbing the larval VNC causes an increase in glial proliferation and a consequent increase in glial cell number ., Although we cannot rule out that other glial cells might also divide , our data demonstrate that this response involves the IG ., We next asked what genes might control the proliferative glial response to injury ., Notch and Pros regulate the mitotic potential of embryonic glia 10; thus , we wondered if they might be involved ., prosvoila1/prosS044116 hypomorphic mutants specifically affected larvae , since embryogenesis proceeded normally but the levels of Pros dropped in IG by the third instar larval stage ( Figure S6A , B ) ., In prosvoila1/prosS044116 VNCs , Ebony was downregulated , meaning that Ebony is a downstream target of Pros ( Figure S6B ) , but there were no major developmental defects as Repo and GS2 expression were normal ( Figure S6C , D ) ., Expression of the Notch antagonist numb with repoGAL4 to knockdown Notch specifically in glia did not cause general developmental defects either ( see below ) ., However , the glial proliferative response to injury was significantly reduced both upon the glial over-expression of numb ( Figure 5A ) and in prosvoila1/prosS044116 mutant larvae ( Figure 5A ) ., In particular , IG number decreased upon stabbing in prosS044116 mutant larvae ( Figure S7 p<0 . 01 ) ., These data show that Notch and Pros are required for the glial proliferative response ., Drosophila Egr/TNF is required for glial proliferation in response to injury in the adult brain 5 , but how it implements this is unknown ., In mammals , TNF can induce cell proliferation via the activation of NFkB 30 , but whether it does in glial progenitors upon CNS injury is unknown ., Thus we asked whether in our injury paradigm , IG proliferation required Egr/TNF , Wengen ( Wgn ) /TNFR , and Drosophila NFκB , Dorsal ., egr/TNF and wgn/TNFR are expressed in the VNC ( Figure S8A–D ) and Dorsal/NFkB is distributed preferentially in Ebony+ pros-lacZ+ IG ( Figure S8E ) ., There were no major developmental defects in the VNC of dorsalH/dorsal1 or egr1/egr3 mutant larvae ( Figure S9 ) ., However , the glial proliferative response was abolished in stabbed egr1 ( unpublished data ) , wgne00637/Df ( 1 ) Exel7463 , and dorsal1/dorsalH mutant larval VNCs ( Figure 5A ) ., These data suggest that Wgn/TNFR , its ligand Egr/TNF , and Dorsal/NFκB are required for the glial proliferative response to injury ., To verify this , we asked whether stabbing resulted in the activation of Dorsal/NFκB in glia ., In its inactive form , Dorsal/NFκB is cytoplasmic , and upon signaling it is translocated to the nucleus to function as a transcription factor ., We found intense nuclear distribution of Dorsal/NFκB in IG upon injury in wild-type ( Figure 5C p<0 . 05 ) , but not in egr1 mutants ( Figure 5C ) ., This shows that stabbing induces the activation of Dorsal/NFκB in IG , which depends on Egr/TNF ( Figure 5B ) ., Therefore , we sought to find out how might Pros , Notch , Eiger/TNF , and Dorsal/NFκB implement their functions in the glial proliferative response to injury ., To investigate the function of Pros in glial proliferation , we generated prosJ013 null mutant MARCM clones in larval glia ., The number of IG in prosJ013 mutant clones ( 1–8 cells per clone in 8 clones generated in n\u200a=\u200a786 VNCs ) did not differ from the number of IG in wild-type clones ( 1–9 cells per clone in 15 clones in n\u200a=\u200a1 , 254 VNCs ) ., Furthermore , in wandering larvae ( 120 h AEL ) the number of glial cells in prosvoila1/prosS044116 mutants was indistinguishable from wild-type ( Figure 6A , D ) ., These data demonstrate that Pros does not affect the extent of glial proliferation in the normal , non-stabbed larva ., However , loss of pros function affected the timing of glial cell division ( Figure S10A , B ) ., In younger ( 96 h AEL ) prosvoila1/prosS044116 mutant larvae , there were more glial cells than in wild-type ( Figure S10A ) , implying that the excess glial cells arose from faster ( but not more ) cell divisions ., Cell division is speeded up by shortening the G1 phase , for instance with the up-regulation of CycE ., Consistently , Pros activates the expression of the CycE repressor Dacapo ( the p21/p27 homologue ) in glia ( Figure S10C p<0 . 05 ) ., To further test if Pros can halt larval glial proliferation , we over-expressed pros in larval glia using tubGAL80ts;repoGAL4 ., This resulted in early larval lethality , and escapers had decreased glial number compared to controls ( Figure 6G ) , showing that Pros inhibits glial proliferation ., Altogether , our data show that Pros functions as a repressor of cycE in glia and it inhibits cell cycle progression by keeping glia arrested in G1 ( Figures 6B ) ., If Pros inhibits cell division , why isnt there glial hyperplasia in the mutants ?, And why cant pros mutant glia proliferate upon injury ?, To solve this conundrum , we wondered if Pros might interact with Notch ., Notch signalling is present in larval IG , its ligand Serrate is in axons ( Figure S11A ) , and Notch maintains the expression of pros ., Pros is also required for Notch signalling ( Figure S11B–G ) , like in embryonic LG 10 ., Thus , Notch and Pros maintain each other in IG ., In other contexts , Notch promotes cell division by regulating the G1/S transition ( Figure S10H ) 43 , 44 ., We found that constitutive activation of Notch signalling—expressing the Notch Intra Cellular Domain ( NotchICD ) —in all glia increased both total glial number ( Figure 6A , Ai , D ) and the number of Ebony+ IG ( Figure 6E ) ., Activation of Notch restricted to the IG only also resulted in an increase in IG cell number ( Figure 6F ) ., Consistently , transient activation of Notch signalling induced PNCA-GFP expression ( Figure S10D , E ) and BrdU incorporation ( Figure S10F , G ) specifically in IG ., These data show that Notch can promote glial cell division ., So if Notch signalling is normally activated in IG , why dont they divide in the intact larva ?, Our data show that Pros and Notch have antagonistic functions in the control of glial proliferation ., Since they also maintain each other , a “tug of war” between Notch and Pros is likely to keep IG in cell cycle arrest ., To test this , we asked whether cell cycle arrest could be evaded by interfering with this feedback loop ., Over-expression of NotchICD in glia resulted in the up-regulation of Pros ( Figure S11E ) , which would repress cycE expression ., When we expressed cycE together with NotchICD , this increased glial number and expanded VNC size ( Figure 6A , D ) ., Over-expressing NotchICD in glia in prosS044116 mutant larvae further increased glial number , causing a tumourous expansion of the VNC ( Figure 6A , Ai , B ) ., The increase in abdominal VNC size was not due to a non-autonomous effect on neuroblast proliferation or increased neuronal number ( Text S2 and Figure S12 and Figure S13 ) , but to increased glial divisions ( Figure 6C ) ., Altogether , our data show that Notch promotes cell cycle progression in glia while Pros inhibits it , and positive feedback between Notch and Pros counterbalances the effects of each other , maintaining glial cells on the brink of dividing ( Figure 6B and Figure S10H ) ., Interfering with this feedback loop has dramatic consequences in glial number and VNC size ., To find out whether Notch and Pros influence IG differentiation , we visualised IG morphology using alrm>mCD8GFP upon loss or gain of function for each of these genes ., To knockdown Notch function only in larvae , we used a temperature sensitive allele of Notch—Notchts1 ., In Notchts1 mutant larvae , IG filopodia and lamellipodia are thinner than in wild-type controls ( Figure 7A ) ., Conversely , over-expression of NotchICD in glia results in larger and rounder glial cells ( Figure 7A ) ., In hypomorphic prosS044116 mutant larvae , IG hardly developed cytoplasmic projections ( Figure 7B ) ., Conversely , over-expression of pros in glia induced more elaborate IG projections ( Figure 7B ) ., These findings show that Notch and Pros have opposite effects on glial differentiation ., To further test how loss of pros function affects IG differentiation , we analysed MARCM clones of prosJ013 null mutant IG: glial morphology was aberrant , with dramatic loss of glial projections compared to wild-type ( Figure 7C , D ) ., The glial differentiation markers Ebony and GS2 were also influenced by Pros ., Ebony is a glial enzyme involved in neurotransmitter recycling 45–48 , and it was down-regulated in prosvoila1/prosS044116 mutants ( Figure 7E and Figure S6B ) ., GS2 is an enzyme involved in Glutamate recycling normally restricted to enwrapping glia 49 ., Larval over-expression of pros with tubGAL80ts; repoGAL4 induced its ectopic expression in non-enwrapping glia ( Figure 7F ) ., Over-expression of NotchICD did not induce Pros , Ebony , or GS2 expression in non-enwrapping glia ( Figure S14 ) ., Thus , GS2 and Ebony are directly regulated by Pros but not by Notch ., Altogether , these findings demonstrate that Pros controls IG differentiation ., We have shown above that the glial proliferative response to injury is abolished in Egr/TNF , Wgn/TNFR , and Dorsal/NFκB mutants ., The number of Repo+ glia , as well as the expression of GS2 and Ebony , were normal in egr1 and dlH/dl1 mutant wandering larvae ( Figure 8A , C and Figure S9 ) , meaning that the glial functions of Egr/TNF and Dorsal/NFκB are dormant in the normal , non-stabbed larva ., To investigate if activation of Dorsal/NFkB could promote glial proliferation , we over-expressed dTRAF2 in all glia ., The Drosophila TRAF6 homologue dTRAF2 binds Wgn/TNFR and induces the nuclear translocation of NFκB homologues 50–52 ., When dTRAF2 was expressed in glia , the number of glial cells increased ( Figure 8A , C ) ., This effect was rescued by expressing dTRAF2 in a dorsalH/dorsal1 mutant background ( Figure 8C ) , showing that the effect of dTRAF2 is mediated by Dorsal ., Activation of Dorsal/NFκB by expressing dTRAF2 in glia resulted in an increased number of Ebony+ IG ( p<0 . 01 ) ., Temporal over-expression of dTRAF2 also induced BrdU incorporation in Ebony+ IG , showing that it activated mitosis cell-autonomously ( Figure S15 ) ., These data show that activation of Dorsal by dTRAF2 promotes glial proliferation ., Glial number and VNC size increased further upon glial expression of dTRAF2 in pros mutants ( UASdTraf2;repoGAL4 prosS044116/prosS044116 Figure 8A , C ) , indicating that Pros antagonizes the proliferative function of Dorsal/NFkB ( Figure 8E ) ., Our data show that Pros-Notch feedback keeps glia on the brink of dividing , and upon injury Egr/TNF signalling via dTRAF2 activates Dorsal/NFκB , tipping the balance towards cell division ( Figure 8E ) ., Thus we asked whether these two genetic mechanisms are linked ., We found that in prosvoila1/prosS044116 mutant larvae Dorsal is decreased from IG ( Figure 8B ) , suggesting that Pros is required for dorsal/NFκB expression ., Since the glial response to injury critically depends on Dorsal/NFkB , this means that the ability of IG to respond to injury is regulated by Pros ( Figure 8E ) ., The glial regenerative response is constrained , as it induces glial proliferation but not tumours , indicating that cell cycle arrest is restored in daughter cells ., Tumorous-like over-growth was induced by expressing dTRAF2 in glia in pros mutants ( Figure 8A , C ) , as was the case when expressing NotchICD in pros mutants ( Figure 6A , Ai , D ) ., This suggested that Dorsal/NFκB might activate pros expression restoring arrest ., To test this , we used hypomorphic prosS044116 mutant larvae that still produce Pros at low levels in a few glial cells ., Expression of dTRAF2 in glia in prosS044116 mutant larvae resulted in the up-regulation of Ebony and Pros ( Figure 8D ) ., These data show that Dorsal/NFκB activates pros expression in glia ., Since Pros inhibits cell cycle progression whereas Dorsal/NFκB promotes it , the “tug of war” between Pros and Dorsal/NFkB is likely to restore G1 arrest in the daughter cells ( Figure 8E ) ., Thus , we have shown that a gene network involving Notch , Pros , TNF , and NFκB controls the balance between glial proliferation , arrest , and differentiation ., To test whether manipulating this gene network was regenerative to enwrapping glia , we examined the glial wound ., Stabbing disrupted the GS2+ Ebony+ glial mesh in the neuropile , and the area devoid of these markers was measured ( Figure 9A ) ., In egr1; prosvoila1/prosS04416 double mutant larvae , in which the proliferative glial response and glial differentiation were both affected , the glial wound increased significantly compared to controls ( Figure 9C ) ., In larvae expressing NotchICD in glia , resulting in over-proliferation , the glial wound was consistently significantly smaller than in controls ( Figure 9A , D ) ., This indicates that either Notch itself or increased glial number is regenerative ., We showed above that over-expression of NotchICD in glia also induced pros expression , and that Pros promoted glial differentiation ., Thus we asked whether the regenerative function of Notch relied on Pros ., When stabbing was carried out in larvae that over-expressed NotchICD in glia but were also mutant for pros ( repoGAL4 prosS044116/UASNotchICDprosS044116 ) , wound size increased significantly ( Figure 9D ) ., Since glial cells proliferated in excess in this genotype ( Figure 6A , D ) , this means that glial proliferation alone is not sufficient for repair and glial differentiation is also required ., To test what consequence the uncovered gene network might have on neuropile repair , we examined cell death levels ., Apoptotic cells were visualized with anti-cleaved-Caspase3 antibodies , and counted in vivo automatically using purposely adapted DeadEasy Caspase software 53 ., Injury increased the extent of apoptosis over non-injured controls ( Figure 9B , E ) ., Expression of NotchICD in glia did not rescue baseline apoptosis , but it rescued injury-induced apoptosis ( Figure 9B , E ) ., This suggests that either NotchICD itself or the resulting increase in glial cell number is protective upon injury ., To test the effect of these genes in enwrapment , we used TEM ., Over-expression of NotchICD in glia dramatically increased glial projections and axonal enwrapment ( Figure 9F ) ., However , enwrapment was reduced when NotchICD was over-expressed in pros mutant larvae ( Figure 9F ) , indicating that Pros is required for enwrapment ., Altogether , these data show that the glial response is regenerative , that both glial proliferation and differentiation are necessary for glial regeneration , and that Notch and Pros play central roles ., To test what effects the glial regenerative response ( GRR ) may have on the axonal bundles , we carried out time-lapse recordings of stabbed larval VNCs , with glial cells labeled with repoGAL4> or alrmGAL4>UAS-DsRed , and all axons labeled with the GFP-protein-trap line G9 ., In wild-type larvae the neuropile wound first increased in size , and numerous vacuoles formed , consistently with TEM and confocal microscopy data from fixed samples ( Figure 10A , G , H and Video S1 , refer also to Figure 2E and Figure 3D , F , G ) ., Subsequently , the vacuoles might disappear and the wound might shrink ( Figure 10B , F , G , H and Video S2 ) ., In Notchts1 mutant larvae , wound size in the axonal neuropile was considerably larger , had greater vacuolization than controls , and did not decrease over time ( Figure 10C , G and Video S5 ) ., Similarly , when glial proliferation was prevented by over-expressing pros in IG , wound size and vacuolization were also more extensive than in controls ( Figure 10D , H and Video S6 ) ., Conversely , when glial cell proliferation was increased by over-expressing NotchICD , wound enlargement and vacuolization were considerably constrained , wound size decreased over time , and even repaired ( Figure 10E , H and Video S7 ) ., In both wild-type and upon over-expression of NotchICD there was a correlation between repair and presence of DsRed+ glial processes within or around the vacuoles and in areas of axonal damage ( Figure 10F and Video S2 , Video S7 ) ., Together with the TEM data ( Figure 3D–N ) , the time-lapse data suggest that upon injury , glial processes engulf the vacuoles , and phagocytose axonal fragments and other cellular debris , contributing to repair ., Altogether , our data show that Notch and Pros control glial proliferation and differentiation required for glial regeneration and debris clearance , and this enables neuropile repair ., Stabbing injury in normal larval VNCs caused an initial loss of IG , wound expansion , and neuropile vacuolization ., Ensheathing glia extended large processes within the neuropile , phagocytosed axonal fragments and cellular debris and dissolved vacuoles , some remaining glial cells divided , and neuropile integrity could be restored ., This natural mechanism was enhanced by activating Notch signalling in glia in the presence of Pros ., Together , NotchICD and Pros prevented wound enlargement and vacuolization , they prevented injury induced apoptosis , increased ensheathing glial number , and promoted glial regeneration and axonal neuropile repair ( Figure 11A ) ., Remarkably , the stabbing injury wound could be completely repaired in these larvae ., This was achieved through the balance of glial proliferation and differentiation under the control of Notch and Pros ( Figure 11B ) ., NotchICD promotes glial proliferation and Pros promotes their differentiation ., In the normal intact larva , the balance between NotchICD and Pros keeps IG in the brink of dividing ., Pros also promotes the expression of cytoplasmic NFκB and of the glial differentiation factors Ebony and GS2 ., Upon injury NFκB shuttles to the nucleus increasing the relative levels of cell cycle activators , and glia divide ., NFκB and NotchICD activate Pros expression , and as Pros levels rise , Pros halts further glial cell division and promotes glial differentiation ., Pros also promotes Notch expression , thus restoring the original balance ., We have shown that interfering with the functions of these genes prevents repair ( Figure 11C ) ., When both glial proliferation and differentiation were inhibited as in egr-pros- double mutant larvae , the glial wound enlarged ., When glial proliferation was abolished in Notchts mutants or upon over-expression of pros in glia , the glial and neuropile w | Introduction, Results, Discussion, Materials and Methods | Organisms are structurally robust , as cells accommodate changes preserving structural integrity and function ., The molecular mechanisms underlying structural robustness and plasticity are poorly understood , but can be investigated by probing how cells respond to injury ., Injury to the CNS induces proliferation of enwrapping glia , leading to axonal re-enwrapment and partial functional recovery ., This glial regenerative response is found across species , and may reflect a common underlying genetic mechanism ., Here , we show that injury to the Drosophila larval CNS induces glial proliferation , and we uncover a gene network controlling this response ., It consists of the mutual maintenance between the cell cycle inhibitor Prospero ( Pros ) and the cell cycle activators Notch and NFκB ., Together they maintain glia in the brink of dividing , they enable glial proliferation following injury , and subsequently they exert negative feedback on cell division restoring cell cycle arrest ., Pros also promotes glial differentiation , resolving vacuolization , enabling debris clearance and axonal enwrapment ., Disruption of this gene network prevents repair and induces tumourigenesis ., Using wound area measurements across genotypes and time-lapse recordings we show that when glial proliferation and glial differentiation are abolished , both the size of the glial wound and neuropile vacuolization increase ., When glial proliferation and differentiation are enabled , glial wound size decreases and injury-induced apoptosis and vacuolization are prevented ., The uncovered gene network promotes regeneration of the glial lesion and neuropile repair ., In the unharmed animal , it is most likely a homeostatic mechanism for structural robustness ., This gene network may be of relevance to mammalian glia to promote repair upon CNS injury or disease . | The process of tissue regeneration has long been studied as a route to understanding what promotes structural robustness of cellular networks in animals ., In the central nervous system ( CNS ) , neurons and glia interact throughout adult life and during learning , at the same time accommodating functional changes while preserving the structural integrity necessary for function ., The mechanisms that confer this combination of structural robustness and functional plasticity in the CNS are unknown , but they may be shared with the cellular responses to injury , which also require structural changes while retaining function ., The glial cells that enwrap axons respond to injury by dividing and re-enwrapping them , leading to partial recovery of function ., Here , we use Drosophila genetics to uncover a gene network underlying this glial regenerative response ., This gene network enables glia to divide upon injury , prevent uncontrolled proliferation , and differentiate ., We find that the network also has homeostatic properties: two cell-cycle activators ( Notch and NFκB ) promote the expression of a cell cycle inhibitor ( Pros ) , providing negative feedback on cell division ., Pros is also essential for glial differentiation , enabling the clearance of cellular debris and axonal enwrapment , and priming glia for further responses ., By removing these genes or adding them in excess , we can shift the response to injury from prevention to promotion of lesion repair ., This gene network is thus a homeostatic mechanism for structural robustness ., Our findings from Drosophila may also help manipulation of glia to repair the damaged human CNS . | cell death, molecular neuroscience, neurobiology of disease and regeneration, mitogenic signaling, animal genetics, cancer genetics, genetic mutation, gene regulation, signaling in selected disciplines, neuroscience, cell differentiation, gene function, animal models, genetics of disease, neuroglial development, developmental biology, drosophila melanogaster, model organisms, molecular cell biology, organism development, developmental signaling, stem cells, cell division, cell growth, molecular development, molecular genetics, adult stem cells, developmental neuroscience, signaling in cellular processes, biology, molecular biology, signal transduction, cellular neuroscience, neural stem cells, cell biology, neurons, genetics, cellular types, evolutionary biology, gene networks, genetics and genomics, evolutionary developmental biology | A gene network involving Notch and Pros underlies the glial regenerative response to injury in the Drosophila central nervous system. |
journal.pntd.0004657 | 2,016 | Treatment Success in Trypanosoma cruzi Infection Is Predicted by Early Changes in Serially Monitored Parasite-Specific T and B Cell Responses | Chagas disease is the highest impact parasitic disease in Latin America and the most common cause of infectious myocarditis in the world 1 ., The goal of treatment of humans in the chronic phase of Trypanosoma cruzi infection is to prevent the development of heart disease and infection by via blood transfusion , congenital transmission and organ transplants 2 ., However , treatment in adult chronic patients is not widely used mainly because of the lack of early metrics of treatment efficacy and the potential adverse effects of these therapeutics 3 ., Several studies in adult patients with mild disease symptoms have demonstrated the clinical benefits of treatment with benznidazole 4 , 5 ., However , the results of the recently published BENEFIT clinical trial 6 has raised questions about the benefits of benznidazole treatment in subjects with established cardiomyopathy , thus emphasizing that therapeutic interventions would have greatest benefit when delivered early in the infection ., The current criterion of a positive response to treatment is the complete loss of reactivity in serially performed conventional serological tests ( ELISA , hemagglutination and immunofluorescence ) , as well as the lack of progression to more severe clinical conditions of Chagas disease ., The decline in serologic titers using current standard tests is very slow , often requiring > 24 months for antibody titers in conventional tests to begin to fall; complete conversion to negative serology can take more than 10 years 4 , 7–11 ., Likewise , disease progression also occurs over decades and does not occur in all infected individuals 4 , 5 ., Consequently , the development of surrogate markers of treatment efficacy is needed for an early assessment of successful treatment and the evaluation of new therapeutic approaches in the chronic phase of T . cruzi infection ., CD4+ and CD8+ T cells derived from patients with chronic T cruzi infection have been shown to produce a variety of cytokines 12–18 ., However recent studies using polychromatic flow cytometry revealed that CD4+ and CD8+ T cells with the capacity to produce only one cytokine ( i . e . monofunctional T cells ) in response to T . cruzi antigens is a common feature in adults with chronic Chagas disease 19–21 ., Of note , monofunctional T cells are more prevalent in patients long-standing infections , generally accompanied by advanced cardiomyopathy 20 , 21 , while polyfunctional T cells are often found in children who have shorter term infections 19 ., This is consistent with the profile of pathogen-specific T cells in other infections where long-term antigen persistence maintains an active pathogen-specific T cell population but with increasing impairment of T cell function over time ., This process known as immune exhaustion has been described for persistent viral , bacterial and protozoan infections 22–27 and is characterized by the loss of IL-2 production , cytokine polyfunctionality , as well as proliferative capacity followed ultimately , by defects in the production of IFN-γ , TNF-α , chemokines and degranulation potential 24 ., Several other features of exhausted T cells , such as high expression of inhibitory receptors , a low expression of the IL-7 receptor and high dependence on the presence of antigen for T cell maintenance have been documented in patients with very long-term T . cruzi infections 20 , 28–30 ., We have proposed that changes in T . cruzi-specific IFN-γ-producing T cells 30 and declines in parasite-specific antibodies as measured by the non-conventional multiplex method might serve as surrogate indicators of treatment success , as determined in a 3-5-year post-treatment follow-up study in chronic Chagas disease patients 7 , 30 ., We hypothesize that treatment decreases parasite load , thus diminishing the antigen necessary to continually activate T . cruzi-specific T cells and B cells ., In patients successfully cured of the infection , a stable change in T and B cell phenotype and activation , in line with antigen-independent immunological memory , would be expected ., In this study , the evolution of the functional profile of T . cruzi-specific T cells and of the humoral immune response to multiple T . cruzi antigens , in association with changes in conventional serological tests — an accepted marker of treatment efficacy — was assessed in 33 subjects chronically infected with T . cruzi over ~8 years following treatment with benznidazole ., We present evidence that cure — assessed by conventional serological tests — achieved many years after treatment with benznidazole was associated with an early decline in T . cruzi-specific IFN-γ-producing T cells , and in antibody titers measured by the multiplex assay ., Changes in the activation status and potential of T . cruzi-specific T cells , indicative of reduced antigen stimulation , provided additional evidence of parasitological cure following benznidazole treatment ., These results further support the case for using immunological markers as indicators of treatment efficacy in T . cruzi infection ., T . cruzi–infected adult volunteers aged 23–54 years were recruited at the Chagas Disease Section of Hospital Interzonal General de Agudos Eva Perón , Buenos Aires , Argentina ., T . cruzi infection was determined by indirect immunofluorescence assay , hemagglutination , and enzyme-linked immunoassay techniques 31 performed at the Instituto Nacional de Parasitologia Dr . Mario Fatala Chaben , Buenos Aires , Argentina ., Chronically infected subjects were evaluated clinically and stratified according to a modified version of Kuschnir grading system 7 , 32 ., Individuals in group 0 had normal electrocardiograph , normal chest radiograph , and normal echocardiograph findings ( n = 27 , median age = 39 years , range = 23–54 years ) , and subjects in group 1 had normal chest radiograph and echocardiograph findings but abnormal electrocardiograph findings ( n = 6 , median age = 42 years , range , 30–50 years ) ., Treatment consisted of benznidazole , 5 mg/kg per day for 30 days 5–9 ., Clinical , serological and immunological analysis was performed prior and after treatment ., Patients enrolled in this study did not change the clinical status during the follow-up period ., This protocol was approved by the institutional review boards of the Hospital Interzonal General de Agudos Eva Perón , Buenos Aires , Argentina and the University of Georgia , GA , USA ., Signed informed consent was obtained from all individuals before inclusion in the study ., PBMCs were isolated by density gradient centrifugation on Ficoll-Hypaque ( Amersham ) and were cryopreserved in a solution of 20% dimethylsulfoxide in heat-inactivated fetal calf serum for later analysis ., Blood to be used for serum analysis was allowed to coagulate at 4°C and centrifuged at 1000 g for 15 min for sera separation ., The number of T . cruzi–specific IFN-γ– and IL-2–secreting T cells was determined by ex vivo ELISPOT using a commercial kit ( ELISPOT Human IFN-γ or IL-2 ELISPOT Set; BD ) , as described elsewhere 33 ., To avoid inter-experiment variations , assays were conducted with paired samples from different time points assayed in the same experiment ., Each time point was assessed 1–3 times ., mAb anti-CD3-fluorescein isothiocyanate ( FITC ) , anti-CD134 ( FITC ) , anti-IFN-γ ( FITC ) , anti-CD25 ( PE ) , anti-CD154 ( PE ) , anti-CD3-peridinin chlorophyll protein ( PerCP ) , anti-CD4 ( PerCP ) , anti-CD27-allophycocyanin ( APC ) , anti-TNF-α ( APC ) and anti-CCR7-phycoerythrin-Cy7 ( PE-Cy7 ) and anti-CD4 ( APC-Cy7 ) were purchased from BD Pharmingen , USA ., PBMCs isolated from T . cruzi-infected subjects were stimulated with 15 μg/ml T . cruzi amastigote lysate or medium alone in 48-well plates at 37°C in a CO2 incubator for 16–20 h ., Ten micrograms of brefeldin A per ml was added to the samples for the last 6 h of incubation ., After stimulation , PBMCs were removed from the plates and stained for cell surface markers followed by fixation and permeabilization with cytofix/cytoperm and intracellular staining with a combination of monoclonal antibodies specific for IFN-γ , TNF-α and CD154 ( CD40L ) ., In order to confirm that cytokine/co-stimulation expression was derived from T cells , antihuman CD3 was added in polyfunctional staining assays in combination with CD4 , IFN-γ and TNF-α or CD4 , IFN-γ and CD154 , respectively ., Typically , 500 , 000 lymphocytes were acquired on a FACScalibur ( Becton Dickinson Immunocytometry Systems , USA ) and analyzed using FlowJo software ( TreeStar , Inc . , USA ) ., Lymphocytes were identified based on their scatter patterns and CD4 expression for the combination of IFN-γ , TNF-α and CD154; and based on scatter patterns as well as CD3 and CD4 expression for the combination of IFN-γ and TNF-α or IFN-γ and CD154 ., Boolean combination gating was then performed to calculate the frequencies of expression profiles corresponding to the seven possible combinations of functions by using FlowJo ., After subtracting the background values , the proportions of the different subsets were expressed as percentages of total cytokine or CD154-positive cells ., Responses to the T . cruzi lysate were considered positive , for any particular subset , if the frequency of cytokine/CD154-positive T cells was threefold higher than the frequency in medium alone and above 0 . 07% of total CD4+ T cells , since the limit of detection was set at 0 . 01% ., Serum specimens were screened for antibodies reactive to a panel of 14 recombinant T . cruzi proteins in a Luminex-based format , as previously described 34 ., Serological responses to each individual T . cruzi protein were considered to have decreased during the study period if the mean fluorescence intensity in at least one recombinant protein declined by 50% relative to that of the time 0 ( pretreatment ) sample assessed concurrently ., Comparisons on the changes in T . cruzi-specific antibodies after treatment , measured by conventional serological tests , were performed using the Mann-Whitney U test ., T cell responses at different time points were compared by Friedman range test ., Comparisons of proportions were performed by use of the χ2 test and Fisher’s exact test ., Differences were considered to be statistically significant at P<0 . 05 ., We have previously shown in a 3–5 year follow-up study that the frequency of peripheral IFN–γ-producing T cells responsive to T . cruzi antigens declined as early as 12 months after treatment with benznidazole and subsequently became undetectable in a proportion of treated subjects 30 ., In some cases , these individuals with declining T cell responses experienced rebounds in parasite-specific T cell responses several years after treatment ., Additionally , some subjects had undetectable IFN-γ-producing T cells ( i . e . below background levels ) prior to treatment that became detectable after treatment , whereas the frequencies of IFN-γ-producing T cells did not change relative to pretreatment in a fourth subset of subjects 30 ., Herein , we report a 4-12-year follow-up ( median 8 years ) of humoral and cellular T cell responses in 33 of these subjects ., All subjects for which IFN-γ ELISPOT responses fell below the level of detection between 12–36 months following treatment with benznidazole ( n = 12 ) showed a later rebound in IFN-γ-producing T cells ( i . e . range 24–72 months post-treatment ) Table 1 , Group 1; Fig 1A ., In contrast , in the remaining subjects , T cell responses did not change significantly during long-term follow up ( Table 1 , Groups 2–4; Fig 1B–1D ) , Likewise , IFN-γ ELISPOT responses are relatively stable in 6 untreated subjects with a 48–60 month-follow-up ( Fig 1E ) ., Monitoring of T . cruzi-specific humoral immune responses assessed by the conventional serological tests , as well as by the multiplex assay that examines responses to 14 individual T . cruzi proteins 34 , was conducted at least yearly following treatment with benznidazole ., The levels of T . cruzi-specific antibodies measured by conventional serology significantly declined over time in subjects with decreased or rebounding IFN-γ-ELISPOT responses following treatment with benznidazole ( Table 2 and Fig 2A and 2B ) whereas antibody titers remained relatively stable in the other patient groups ( Table 2 , Fig 2C and 2D ) ., Of note the seven patients who showed conversion from seropositive to seronegative–the standard metric of infection cure—on at least 2 of the 3 conventional serological tests were patient groups 1 and 2 ( Table 2 , Fig 2A and 2B ) ., Conversion from seropositive to seronegative was observed on average >5 years post-treatment ( 24–96 months ) and was sustained up to 12 years post treatment ( Fig 2B , subject PP31 ) ., In concordance with conventional serology , a multiplex assay utilizing recombinant proteins from T . cruzi also revealed a higher rate of declining antibody titers among subjects with decreased or rebounding ELISPOT responses ( Table 2 , Fig 3A–3D ) ., Seventeen out of nineteen patients with a rebound or a significant decrease in IFN-γ-producing T cells following treatment with benznidazole showed a fall in the levels of antibodies specific for one more recombinant proteins in comparison to 4 out of 13 in the group of patients in which T cell responses remained unchanged or became detectable after treatment ( Table 2 , Fig 3A–3D ) ., Notably , the multiplex assay detected declines in antibody levels as early as 2–24 months post-treatment ( Fig 3A–3D ) while declines in conventional serologic tests were not evident until 24–48 months post-treatment ( Fig 2A and 2B ) ., Conversion from seropositive to seronegative by conventional serological tests can take up to 9 years to occur ( Fig 2A and 2B ) ., Thus , declines in T . cruzi-responsive IFN-γ-producing T cells and T . cruzi-specific multiplex-detected antibodies following benznidazole treatment preceded and were predictive of conversion to negative conventional serology , the accepted standard of treatment success ., As previously reported 30 , IL-2-producing T cells were low in chronically T: cruzi-infected subjects and changed in concert with IFN-γ T cell responses after treatment with benznidazole ( Fig 2A–2E ) ., Treatment success as measured by declining T . cruzi-specific antibody responses was not associated either with the age of subject at initiation of treatment or the baseline T . cruzi-specific antibody titers ., However , subjects with declining antibody titers as a group had higher pre-treatment frequencies of IFN-γ- and IL-2 producing T cells as compared to patients who showed no change in humoral responses following treatment ( Fig 4 ) ., Since rebound in T . cruzi-specific T cells making IFN-γ was associated with declining serological titers , suggestive of a decreased presence of parasite antigen , we hypothesized that these T . cruzi-responsive T cells re-emerging long-term after treatment would result in enhanced functional capacity of T . cruzi-specific T cells ., Group 1 subjects exhibited an increase in single CD4+CD54+ and CD4+IFN-γ+ T cells ( Fig 5B–5D ) coincident with a decrease in single CD4+TNF+ T cells ( Fig 5B–5E ) following treatment with benznidazole ., Some subjects also showed an increase in dual IFN-γ+CD154+ T cells ( Fig 5C–5E ) or polyfunctional T cells with the ability to express IFN-γ; TNF-α; and CD154 ( Fig 5E ) , These findings show that successful treatment resulted in a change of the functional profile of parasite-specific T cells with a restoration of the co-stimulatory function , generally impaired in chronic infections ., One of the primary drawbacks in treatment of chronic T . cruzi infections is the difficulty of assessing treatment efficacy 4 , 35 , 36 , principally in the short term ., In this study , we investigated if the early , post-treatment changes in T . cruzi-specific T cell and antibody responses , previously reported by our group 30 , are predictors of treatment efficacy ., To answer this question we compared these non-conventional immune assessments with the conversion from positive to negative conventional serology — the accepted standard of cure — in a longitudinal over ~8-year post-benznidazole treatment follow-up study ., Our study revealed that cure—as determined by seronegative conversion by conventional serology—was strongly correlated with an early decline in both T . cruzi-specific T cells and in the levels of antibodies specific for a panel of T . cruzi antigens ., Significant declines in IFN-γ-producing T cells and multiplex-monitored antibody responses post-treatment also preceded detection of reductions in anti-T ., cruzi antibodies detectable by conventional serological tests ., In contrast , subjects exhibiting stable T cell responses post-treatment were generally associated with unaltered conventional and multiplex-assessed humoral responses ., Thus , this work identifies dependable and early markers of treatment efficacy in Chagas disease ., These results support and extend our previous studies 7 indicating the superiority of assaying responses to >10 recombinant proteins using a multiplex format over conventional serologic tests ., Other studies have also demonstrated that the use of recombinant proteins as antigens can often detect changes in parasite-specific antibodies earlier than the complex T . cruzi antigen preparations normally used in many conventional tests 37 , 38 ., However , in 15 out of the 33 patients evaluated in this study slight or no changes in T . cruzi-specific humoral and cellular T cell responses were observed , suggesting a failure of treatment and confirming previous studies showing that benznidazole treatment is not uniformly successful curing T . cruzi infection 4 , 6 ., Some subjects with declining or negative anti-T ., cruzi antibody levels and T cell responses experienced rebounds in T cell responses , prompting the question of whether these T cell reflected renewed antigen stimulation , and thus persistence of T . cruzi infection ., However rebounding IFN-γ-producing T cells were associated with decreasing serological titers by both conventional and multiplex assays and two of the seven subjects who converted to negative conventional serology–the accepted standard of cure—exhibited this rebound in T cell responses ., Therefore , it seems likely that these parasite-specific T cells in rebound responses are maintained in the absence of or very low levels of antigen , a characteristic of TCM ., Such responses are evident in mice cured of T . cruzi infection by benznidazole treatment 8 , 39 , 40 ., Herein , benznidazole treatment resulted in a different functional quality of CD4+ T cells with a prominent decline in single producers of TNF-α and an increase in either monofunctional or polyfunctional CD4+ T cells expressing CD154 after treatment ., Several studies have shown that constant antigen stimulation during chronic infections might skew T cell responses to single TNF-α-producing T cells 41 and low CD154 expression 42 , 43 which are restored after suppression of antigen load 41 , 44 ., Other studies have also shown that therapy with benznidazole in the chronic phase of the infection resulted in a shift toward a type- T cell profile profile 45–47 ., Collectively , these findings further support that parasite persistence in chronic T . cruzi infection induces significant alterations in T cell function ., An interesting observation that deserves further investigation is that subjects who showed the greatest decrease in T . cruzi-specific antibodies following treatment also had on average higher baseline levels of IFN-γ-producing T cells compared with subjects with modest or no changes in humoral responses ., Studies in the experimental models have suggested that the quality of the anti-T ., cruzi immune response plays a role in the efficacy of benznidazole treatment 48–51 ., Studies in larger patient groups and in experimental models are needed to confirm these findings ., This study validates the ability of appropriate and sensitive immunological tests to provide early evidence of treatment efficacy in chronic Chagas disease ., Providing tools to not only monitor but to more rapidly predict treatment success or failure will facilitate the development of new and better therapeutic options in Chagas disease . | Introduction, Methods, Results, Discussion | Chagas disease is the highest impact parasitic disease in Latin America ., We have proposed that changes in Trypanosoma cruzi-specific immune responses might serve as surrogate indicators of treatment success ., Herein , we addressed in a long-term follow-up study whether cure achieved after treatment can be predicted by changes in non-conventional indexes of anti-parasite serological and T cell activities ., T . cruzi-specific T cell responses , as measured by interferon-γ ELISPOT and T . cruzi-specific antibodies assessed by ELISA , hemagglutination and immunofluorescence tests as well as by a multiplex assay incorporating 14 recombinant T . cruzi proteins were measured in 33 patients at 48–150 months post-benznidazole treatment ., Cure — as assessed by conventional serological tests — was associated with an early decline in T . cruzi-specific IFN-γ-producing T cells and in antibody titers measured by the multiplex serological assay ., Changes in the functional status and potential of T . cruzi-specific T cells , indicative of reduced antigen stimulation , provided further evidence of parasitological cure following benznidazole treatment ., Patients showing a significant reduction in T . cruzi-specific antibodies had higher pre-therapy levels of T . cruzi-specific IFN-γ- producing T cells compared to those with unaltered humoral responses post-treatment ., Monitoring of appropriate immunological responses can provide earlier and robust measures of treatment success in T . cruzi infection . | This study demonstrates that alterations in immunological parameters early after treatment with benznidazole in Chagas disease patients are predictors of treatment efficacy ., Cure was associated with an early decline in T . cruzi-specific IFN-γ-producing T cells and in antibody titers and with high basal levels of T . cruzi-specific T cells . | blood cells, medicine and health sciences, immune cells, enzyme-linked immunoassays, pathology and laboratory medicine, immunology, tropical diseases, parasitic diseases, parasitic protozoans, protozoans, neglected tropical diseases, immunologic techniques, antibody response, research and analysis methods, white blood cells, serology, animal cells, proteins, t cells, immunoassays, recombinant proteins, protozoan infections, immune response, trypanosoma cruzi, biochemistry, trypanosoma, chagas disease, cell biology, biology and life sciences, cellular types, organisms | null |
journal.pcbi.1006381 | 2,018 | Identification of excitatory-inhibitory links and network topology in large-scale neuronal assemblies from multi-electrode recordings | Understanding the relationships between structure and function , dynamics and connectivity of neuronal circuits are a challenge of the modern neurosciences , especially as the characterization of neuronal interaction in terms of functional and effective connectivity 1–3 is concerned ., Functional connectivity is an observable phenomenon defined as statistical dependency between remote neurophysiological events; it is usually inferred on the basis of correlations among neuronal activity measurements , by means of different approaches ranging from basic cross-correlation4 to model-based methods1 , 5 ., Effective connectivity refers explicitly to the influence that a neuron or neural system exerts on another one , either at synaptic or population level; it can be inferred by perturbing the activity of a neuron , and then by measuring the other neurons activity changes . Structural or anatomical connectivity is related to the physical connections ( i . e . , synapses ) among neurons 2 ., In this paper , we refer to the more general framework of functional connectivity , even if , by using the proposed correlation-based method , directed graphs ( i . e . causal relationships ) can be derived ( cf . Materials and Methods sect . ) ., The complexity of the nervous system and the difficulties of multi-site parallel recordings in in vivo experimental models , hampered the systematic study of emergent properties of complex networks ., At the same time , the availability of validated methods able of reliably inferring functional connections down to synaptic level is still limited ., To this end , we adopted a reductionist approach making use of in vitro experimental models coupled to Micro-Electrode Arrays ( MEAs ) ., In this context , large-scale neural networks developing ex vivo and chronically coupled to MEAs 6 , represent a well-established experimental system for studying the neuronal dynamics at population level 7 ., Despite their simplicity , they show recurrent synchronized periods of activity , as also observed in vivo during sleep or anesthesia , and even quiet wakefulness 8 , 9 ., These model systems represent a good trade-off between controllability-observability and similarity to the in vivo counterpart , allowing accessibility and manipulation from both chemical and electrical point of view ., Recent advances in multichannel recording techniques have made possible to observe the activities of thousands of neurons simultaneously with the acquisition of massive amount of empirical data 10 ., These methods are very attractive since they allow the detailed monitoring of the on-going electrophysiological spatio-temporal patterns of complex networks 11–14 ., Reconstructing the detailed functional connectivity of a neuronal network from spikes data is not trivial , and it is still an open issue , due to the complexities introduced by neuron dynamics and high anatomical interconnectivity 15 , 16 ., Statistical analysis of spike trains was pioneered by Perkel 17 and followed by more than four decades of methodology development in this area 18 ., Cross-correlation based methods remain the main statistics to evaluate interactions among the elements in a neuronal network , and produce a weighted assessment of the connections strength ., Weak and non-significant connections may tend to obscure the relevant network topology made up of strong and significant links , and therefore they are often discarded by applying an absolute or a proportionally weighted threshold 19 ., Correlation-based techniques include independent components analysis , synchrony measures 20 , cross-correlation 21 , 22 , correlation coefficients 7 , 23 , partial-correlation 24 ., Other widespread techniques to infer functional connectivity are based on Information Theory ( IT ) methods 10 , 25 , 26 , Granger causality 27 , 28 and dynamical causal modeling 1 ., With few exceptions 29 , 30 , all the recently introduced and revisited methods concentrate on excitation , ignoring inhibition or admitting the failure in reliably identifying inhibitory links 26 ., In this work , we focus attention on cross-correlation histogram ( CCH ) based methods ., We present a new algorithm able to efficiently and accurately infer functional excitatory and inhibitory links; we validate the method on simulated neuronal networks; finally , we study connection properties in large-scale ex vivo neuronal networks showing how to directly and reliably derive the topological properties of such networks ., There are three different connectivity conditions that , theoretically , influence the temporal correlation between neurons: pairs of excitatory neurons , pairs of inhibitory neurons , and inhibitory-excitatory pairs 31 ., The first term is the one usually estimated and from which we obtain the inferred functional excitatory network usually represented by a ( directed ) graph ., The second term is negligible as inhibitory-inhibitory links are physiologically very sparse 32 ., The last term , when it is exerted by a GABAergic interneuron to cortical excitatory neurons , acts by reducing the activity and decreasing the spontaneous fluctuations ( i . e . , feedforward inhibition ) ., On the contrary , when it is exerted by cortical excitatory neurons to GABAergic interneurons , it acts by increasing the activity of such neurons that , in turn , form inhibitory synaptic contacts with the glutamatergic cortical cells ( i . e . , feed-back inhibition ) 33 ., In other studies 34–36 , it was noticed the primary effect of inhibition is a trough in the cross-correlogram: to detect this interaction a background of postsynaptic spiking against which the inhibitory effect may be exercised ( i . e . , high and tonic firing rates ) is needed 22 ., From experimental works related to in vivo multi-unit recordings , it was shown the sensitivity to excitation is much higher than the sensitivity to inhibition 37 ( due to the low firing rates of neurons ) ., Finally , it should be underlined the analysis of interactions in neuronal networks is a quite demanding computational process , and all the currently proposed methods for analyzing multiple spike trains rely on quantities that need to be computed through intensive calculations 38 ., By using the ad-hoc developed CCH , we could derive functional connectivity maps ( both for excitation and inhibition ) and to reliably extract topological characteristics from multiple spike trains in large-scale networks ( i . e . , thousands of neurons ) monitored by large-scale MEAs ( i . e . , thousands of micro-transducers ) ., Starting from the standard definition of the cross-correlation 22 ( cf . , Materials and Methods sect . ) , we adopted the normalization approach described in 21 , 39 to obtain the “raw” Normalized Cross-Correlation Histogram ( NCCH ) ., We formalized our hypothesis that , the extraction of negative peaks ( rather than troughs ) obtained by a filtering operation on the NCCH and followed by distinct thresholding operations for excitatory and inhibitory connections allows to identify a significant percentage of inhibitory connections with a high-level accuracy at low computational cost ., Theoretically , cross-correlation is able to detect both an increase and a decrease of the synchrony between spike trains related to putative interconnected neurons ., However , in real experimental data , the cross-correlogram is very jagged making difficult the detection of small peaks and troughs , and , apart from specific conditions ( i . e . , high and tonic firing rate ) 4 , hindering the detection of inhibition ., Our approach consists in a simple post processing of the cross-correlation histogram , thus obtaining what we called Filtered and Normalized Cross-Correlation Histogram ( FNCCH , curly brackets in Eq ( 1 ) ) ., Stated a reference neuron x and a target neuron y , Eq ( 1 ) provides the mathematical definition of the absolute peak of the FNCCH ., FNCCHxypeak=Cxy ( τ ) |=argmaxt{|Cxy ( t ) −1W∑v=−W2v=W2Cxy ( v ) |}, ( 1 ), where W is the time window where FNCCH is evaluated ., The filtering procedure ( cf . Materials and Methods sect . ) consists in subtracting the mean value of the cross-correlogram ( in the time window W ) from the values of the normalized cross-correlogram Cxy ( ν ) , ν ∈ -W/2 , W/2 ., The subsequent peaks extraction operation is performed by considering the absolute values , and it allows to compute the highest peak ., In this way , it is possible to distinguish between peaks and troughs by taking into account the original signs: a positive value refers to an excitatory link , and a negative value refers to an inhibitory one ., Details about further refinements needed to avoid detection of false inhibitory connections can be found in the Supplementary Information ( cf . , Sect . S1 ) ., In the next sections , we show the validation of the method with the aid of large-scale in silico networks; then , we present the results , in terms of functional connectivity maps and network topology , obtained from the analysis of multi-electrode parallel recordings of in vitro neuronal populations ., Such populations are coupled to both 60 channels MEAs ( MEA-60 ) and high-density MEAs with 4096 micro-transducers ( MEA-4k ) ( cf . Materials and Methods sect . ) ., We applied the FNCCH ( time window W = 25 ms and time bin 1 . 0 ms ) to 10 realizations of in silico neural networks made up of 1000 randomly connected neurons , characterized by an average ratio between inhibitory and excitatory connections of 1/4 ( cf . , Materials and Methods sect . ) ., The model was tuned to reproduce the dynamics exhibited by in vitro neuronal networks ., Simulations show the typical signature characterized by a mix of spiking and bursting activities as displayed by the raster plot and the Instantaneous Firing Rate ( IFR ) traces of the excitatory ( red ) and inhibitory ( blue ) neuronal populations of Fig 1A ., From a topological point of view , both the excitatory and inhibitory structural sub-networks follow a random connectivity , as the incoming degree distributions of Fig 1B ( inset ) display ., Each neuron receives 100 connections from the other neurons: excitatory neurons receive 80% of excitatory and 20% of inhibitory links , respectively , ( reflecting the ratio of the excitatory and inhibitory populations ) ; inhibitory neurons receive only excitatory connections ( cf . S2C Fig ) ., Further details about the dynamics and connectivity of the simulated neuronal networks can be found in the Supplementary Information ( cf . , Sect . S2 ) ., Fig 1C and 1D quantify the performances of the FNCCH by means of the Receiver-Operating-Characteristic ( ROC ) 40 curve and the Matthews Correlation Coefficient ( MCC ) 41 ., Fig 1C shows the ROC curves obtained by comparing the Synaptic Weight Matrix ( SWM ) of the model ( i . e . , the ground truth ) with the computed Functional Connectivity Matrix ( FCM ) , and Fig 1D shows the MCC curve ( cf . , Materials and Methods sect . ) ., The ROC curve relative to the detection of inhibitory connections ( blue curve in Fig 1C ) is very close to the perfect classifier , with an Area Under Curve ( AUC ) of 0 . 98 ± 0 . 01 ( blue bar in the inset of Fig 1C ) ., The MCC curve relative to the inhibitory links ( blue curve in Fig 1D ) has a maximum value of 0 . 87 ± 0 . 04 , showing a good precision in the identification of inhibition ., Then , we compared the sensitivity of the FNCCH for the detection of excitatory links ( red curves in Fig 1C and 1D ) with the standard NCCH’s one ( for excitation , black curves in Fig 1C and 1D ) to underline the improved detection capabilities obtained by the filtering procedure ., We observed not only a significant ( p < 0 . 001 ) AUC increase ( 0 . 92 ± 0 . 01 vs . 0 . 72 ± 0 . 02 , Fig 1C inset ) , but also significant improvements in both ROC and MCC curve shapes for low values of false positive rates ( FPR ) ., In particular , we can notice ( Fig 1D ) , that the FNCCH excitatory curve has a maximum value of about 0 . 75 with respect to the correspondent NCCH value ( for the same false positive rate ) that is negative ( suggesting a disagreement between prediction and observation ) ., Further details about false and true positive detection can be found in the Supplementary Information ( Sect . S5 ) ., The above results justify the use of a hard threshold procedure ( cf . , Materials and Methods sect . ) to select the strongest and significant functional connections ., The Thresholded Connectivity Matrix ( TCM ) is thus directly computed from the FCM by using a threshold equal to ( μ + 1 σ ) , ( mean plus one standard deviation of the connections strength ) for the inhibitory links , and ( μ + 2 σ ) for the excitatory ones , obtaining estimated links with a very high-level of accuracy ( cf . Materials and Methods sect . ) : R2 = 0 . 99 for the inhibitory links and R2 = 0 . 94 for the excitatory ones ., To investigate whether the reconstructed functional connectivity network resembles the one of the model , we calculated the excitatory and the inhibitory ( Fig 1B ) links degree distribution after the thresholding procedure from TCM ., The computed degree-distributions fit a Gaussian distribution ( Fig 1B , R2 = 0 . 99 for the inhibitory links and R2 = 0 . 98 for the excitatory ones ) , in accordance with the original distributions used to generate the structural ( random ) connectivity of the model ( Fig 1B inset ) ., It can be noticed that the mean and standard deviation values of the functional Gaussian distribution for the excitatory links are in good agreement with the structural ones ( μfunct = 87 , σfunct = 13 . 2 and μstruct = 80 , σstruct = 19 . 6 ) ., On the other hand , for the inhibitory links , such values are higher than the structural ones due to the presence of many polysynaptic interactions ( μfunct = 48 , σfunct = 9 . 3 and μstruct = 25 , σstruct = 14 . 5 ) ., Finally , we computed the delay distribution for both the excitatory and the inhibitory links from the TCM ( Fig 1E ) ., The extracted delay distribution for the excitatory links qualitatively reflects the one used to generate the model ( uniform distribution in the interval 0 , 20 ms ) ., The estimated inhibitory distribution , instead , exhibits a more confined range which reflects the one used to produce the model ( constant delay set at 1 ms ) , but with a spread and a median value at about 5 ms ( cf . , Materials and Methods sect . ) ., The disagreement can be explained by the presence of multiple and polysynaptic interactions ( due to the combination of excitatory and inhibitory inputs on a single neuron; cf . , Discussion sect . ) ., Further validation of the proposed method was pursued by implementing a scale-free ( with small-world features ) network ., The results ( cf . Supplementary Information , S3 Fig ) are less striking than those obtained for random connectivity; nevertheless , FNCCH outperforms standard cross-correlation and the identification of inhibitory links is still maintained with a similar general trend ., The FNCCH was applied to neuronal networks coupled to two different devices: MEA-60 and MEA-4k ., Fig 2 shows the two utilized microtransducers ( Fig 2A and 2D ) and illustrative images of networks coupled to the two ( Fig 2B and 2E ) ., Such networks are the morphological substrate originating the complex electrophysiological activity characterized by an extensive bursting dynamics ( i . e . , highly synchronized network bursts ) and a random spiking activity ., Fig 2C and 2F show two examples of spontaneous activities recorded by a MEA-60 ( Fig 2C ) and a MEA-4k ( Fig 2F ) ., We can observe silent periods , desynchronized spiking activity , and peaks of activity ( of different duration and called network bursts ) , which cause a rapid increase of the Instantaneous Firing Rate ( IFR ) ( Fig 2C and 2F , bottom panels ) ., More details about the spiking and bursting dynamics originated by networks coupled to MEA-4k are reported in the Supplementary Information ( S1 Table ) ., We analyzed three cortical and three striatal networks coupled to the MEA-60 ( FNCCH parameters: time window W = 25 ms and time bin 0 . 1 ms ) and three cortical networks coupled to the MEA-4k ( FNCCH parameters: time windows W = 24 ms and time bin of 0 . 12 ms ) after they reached a stable stage ( i . e . , after 21 Days In Vitro , 21 DIV ) ., Fig 3A and 3G show connectivity graphs of cortical and striatal networks coupled to a MEA-60 device ( Fig 3B and 3H and 3C and 3I show the contribution of excitation and inhibition , respectively ) ., All the graphs were obtained by applying the hard threshold approach and the spatio-temporal filtering to prune co-activations ( cf . , Materials and Methods sect . ) ., Then , we looked , for the cortical networks , the presence of privileged sub-networks constituted by the most connected nodes ( i . e . , rich club ) , by computing the Rich Club Coefficient ( RCC ) curve 42 ( cf ., , Materials and Methods sect ., , Eq ( 10 ) ) ., The nodes of these sub networks are highlighted in yellow and cyan ( Fig 3B and 3C ) ., For the striatal culture , the qualitative prevalence of inhibitory connections is clearly visible ., To characterize the detected links for the cortical cultures , we computed the box plots of the functional connection peak delays ( Fig 3D ) and lengths ( Fig 3E ) of the excitatory ( red ) and inhibitory ( blue ) connections ., Similar graphs derived from a cortical network coupled to a MEA-4k were obtained ( Fig 4A ) ., Links strength is represented by two color codes ( arbitrary unit ) for excitation ( hot-red color code ) and inhibition ( cold-blue color code ) ., The two detected subnetworks are also shown in Fig 4B and 4C ., Moreover , the box plots pointing out the connection peak delays and lengths are depicted in Fig 4F and 4G ., Noteworthy it is that the inhibitory links are slower , and with possible slightly longer connections than the excitatory ones , as reported in literature for structural and functional connectivity in brain slices 43 ., Considering the high number of connections found by using the MEA-4k , we point out the two hundred strongest connections for excitation and the fifty strongest connections for inhibition ( Fig 4D and 4E ) , illustrating how these main links include both short and long interactions with a prevalence of short interactions for excitatory connections ., We also computed the inhibitory links percentage with respect to the total number of detected links for the three different experimental conditions and three experiments for each condition ., As expected , we found that striatal cultures have a higher percentage of inhibition and inhibitory links ( about 60% ) 44 , 45 than cortical ones ( about 25% ) ., It is worth noticing that for the cortical cultures the excitatory/inhibitory ratio is detected quite independently of the number of recording sites ( Figs 3F and 4H ) , although it tends to stabilize with a shorter recording time for the MEA-4k ., Interestingly enough , the found ratio ( about 1/4 ) in cortical networks between inhibitory and excitatory links is roughly the same as the ratio of inhibitory and excitatory neurons as estimated by immunostaining in similar experimental preparations 8 ., In order to derive the topological features 46 of the analyzed cortical networks , we computed the Clustering Coefficient , CC ( Fig 5A ) and the average shortest Path Length , PL ( Fig 5B ) ., Then , we extracted the Small-World Index ( SWI ) by comparing the CC and the PL of the analyzed networks with the mean values of CC and PL of 100 realizations of a random network with the same degree-distribution , as recently proposed 26 ., We found that when cortical networks are coupled to MEA-4ks devices , we can see the emergence of a clear small-world ( SW ) topology ( Fig 5C ) ; on the other hand , for cortical networks coupled to MEA-60s devices , we cannot infer any SW topology ., From the measurements performed by MEA-4ks , we can state that both inhibitory and excitatory subnetworks with their small world index , SWI >>1 ( 9 . 2 ± 3 . 5 for the inhibitory links and 5 . 2 ± 2 for the excitatory ones ) contribute to ‘segregation’ ., Moreover , both inhibitory and excitatory links with their fraction of long connections contribute also to network ‘integration’ ( i . e . , communication among the SWs ) ., To further characterize the topology of these neuronal assemblies , we also investigated the possible emergence of scale-free topologies 47 by evaluating the presence of hubs48 and power laws for the excitatory ( Fig 5D ) , inhibitory ( Fig 5E ) and global ( Fig 5E , inset ) link degree distributions ., In agreement with previously published model systems 49 and other studies 43 , we obtained that such distributions fit a power law with R2 higher than 0 . 92 , in all the three cases ., Finally , we searched for the presence of privileged sub-networks made up of the most connected nodes ( i . e . , rich club ) of the investigated networks by computing the RCC curve ., For the analyzed cortical cultures , we found privileged sub-networks as indicated by the computed RCC values with a maximum value of 2 . 7 ± 0 . 5 ., Fig 4B and 4C show the rich club networks identified for one neural network coupled to the MEA-4k , represented by means of blue circles ( for excitatory subnetwork ) and pink circles ( for inhibitory subnetworks ) ., Fig 3B and 3C are the analogous for a cortical neural network coupled the MEA-60 ( yellow for the excitatory nodes and light blue for the inhibitory ones ) ., Similar cortical networks coupled to the MEA-60 devices show no clear SW topology ( Fig 5C ) ; these networks seem to be characterized by a ( sub ) -random topology with SWI of 0 . 4 ± 0 . 1 for the excitatory and 0 . 2 ± 0 . 2 for the inhibitory links ., These cortical networks are of the same type as the ones coupled to the MEA-4k ( i . e . , similar density of neurons , same age , same culture medium ) , and the apparent estimated random topology should be attributed to the low number of recording sites ( i . e . , 60 channels ) that are not enough to reliably infer topological features ., To determine how the number and density of electrodes are crucial , we computed the SWI by considering a reduced number of electrodes for the functional connectivity analysis from the MEA-4k recording , as described in Fig 5F ., In particular , we started from the full resolution of the MEA-4k ( i . e . , 4096 electrodes ) , and we progressively decreased the electrode density to 60 electrodes ( inter-electrode distance 189 μm , electrode density 19 electrode/mm2 ) to obtain a configuration comparable with the MEA-60 devices , as previously reported50 ., The obtained results are shown in Fig 5G: the SWI decreases down to a random topology becoming variable and unstable when the number of the considered electrodes is less than 100 ., This last result is referred to the excitatory links and the same analysis was not applied to the inhibitory connections ., Such inhibitory links are much less than the excitatory ones , thus leading to an inhibitory topology reconstruction that is strongly influenced by the decimation scheme applied to reduce the number of electrodes ., Generally , by inspecting a CCH , we can notice an increase or a decrease of the fluctuations 22 ., In some studies , it was noticed that the primary effect of inhibition on the cross-correlogram is a trough near the origin , and for this interaction to be visible there must be present a background of postsynaptic spiking against which the inhibitory effect may be exercised ( high-tonic firing rate regime ) 4 , 35 ., From experimental works related to the analysis of connectivity from cortical multi-unit recordings 55 , a good sensitivity for excitation is obtained , while the situation is considerably worse for inhibition ., This is due to a low sensitivity of CCH for inhibition , especially under the condition of low firing rates 4 , 56 ., The difference in sensitivity may amount to an order of magnitude , and it was demonstrated that for inhibition , the magnitude of the departure relative to the flat background is equal to the strength of the connection , whereas for excitation it involves an additional gain factor 4 ., As a whole , the lack of efficiency in the detection of inhibition , simply reflects the disproportionate sensitivity of the analysis tool 57 ., In our work , we introduced a cross-correlogram filtering approach ( FNCCH ) developed to overcome the inhibition detectability issue ., As Fig 1 shows , the FNCCH is able to detect , with high accuracy , the inhibitory links when applied to in silico neural networks with similar dynamics with respect to the actual ones ., The filtering procedure improves also the detectability of the excitatory links , resulting in a reshaping of the ROC curve ( Fig 1A ) with an increase of both precision ( MCC curve , Fig 1B ) and AUC with respect to the standard cross-correlation ( NCCH ) ., However , the presented FNCCH , being a CC-based method , has some limitations in the inhibitory links detection that we tried to investigate with our in silico models ., The main factor affecting the detectability of inhibition , is the variability of CC ., In order to reduce this variability , it is possible to increase the number of coincidences per bin by widening the bin-width ( that is , down-sampling with loss of information in the acquired electrophysiological data ) , or by increasing the number of involved events ( which can be obtained with high firing rate and/or by raising the recording time ) 58 ., Another influencing factor depends on the balance of excitatory and inhibitory neuronal inputs ( i . e . , balanced model ) and it is referred to the relative strength between inhibitory and excitatory inputs ., In fact , when the neuron is not balanced , excitation is , on average , stronger than inhibition ., Conversely , when the neuron is balanced , both excitation and inhibition are strong and detection of inhibitory links improves 22 , 31 , 57 ., Starting from the in silico model , we were able to investigate the impact of rates variability on excitation/inhibition detectability , and to try to define a reasonable threshold ( criterion for detectability 22 , 56 ) ., In particular , we varied the firing rate of the inhibitory neurons from 20 spikes/s to 2 spikes/s , while maintaining a firing rate of 2–3 spikes/s for the excitatory neurons ., We found that the detectability of the functional inhibitory links is preserved with our method , down to a firing of about 6 spikes/s , and then decreases significantly ., We investigated also the inhibition identification with respect to the recording time ., Starting from 1 hour of simulation , we reduced ( 10 min steps ) the recording time , and we found that there is a decrease in the inhibition detectability below 30 minutes of recording ( cf . S4 Fig ) ., Finally , we investigated the performances of the FNCCH in a scale-free and small-world network ., The detection of inhibition was still possible with relatively good results , even if the global performances of the algorithm decreases ., This shall not be attributed to the scale-free topology , but to the reduced firing rate for both inhibitory and excitatory neurons and to possible unbalances between inhibition and excitation ( cf . S3 Fig ) ., Nevertheless , the method could reliably capture the topology of the network and qualitatively estimate the synaptic in-degree distribution ., Thus , the obtained results enabled us to apply the FNCCH to in vitro large-scale neural networks , and allowed us to infer topology and functional organization ., The described procedure could be also directly applied to Multi Unit Activity ( MUA ) from in vivo multi-site measurement recordings ., Other methods ( e . g . , partial correlation , transfer entropy ) were not taken explicitly into consideration for comparison , either for their computational costs , or for the inability to identify inhibitory links 59 ., The cortical networks probed with MEA-4k showed a clear small-world topology ., The inhibitory functional links had a SWI equal to 9 . 2 ± 3 . 5 , higher than the value extracted from the excitatory links ( 5 . 1 ± 1 . 9 ) ., Conversely , the cortical networks coupled to the MEA-60 showed a random organization topology ( 0 . 21 ± 0 . 212 for the inhibitory links and 0 . 38 ± 0 . 1 for the excitatory ones ) ., These apparent random organizations are due to the low number of recording sites of the acquisition system; in fact , it is worth to remember that the SWI is computed by comparing cluster coefficient ( CC ) and average shortest path length ( PL ) of the analyzed networks to the corresponding values for surrogate random equivalent networks ( same number of nodes and links ) ., From the obtained results , unlike recently presented findings 42 , we demonstrated that the emergence of small-worldness , cannot be reliably derived or observed in a neuronal population probed by a reduced number ( < 100 ) of recording sites ., To characterize connectivity properties , besides the importance of well-defined statistical tools used for the analysis , it is fundamental to probe network activity by using large-scale microtransducer arrays ( i . e . , with at least 200 electrodes ) ., As a whole , the issue related to the low number of recording sites should be carefully taken into account when extracting dynamical features as well as organizational principles of complex networks ., Finally , it should be underlined that we focused our attention on CC based methods ., We mentioned , in the Introduction , the widespread use of Information Theory ( IT ) based techniques ., Beside the relative novelties of such methods , and the good performances ( for a review see 38 and references therein ) , they showed high computational costs and , to our knowledge , the inability to reliably estimate inhibitory connections 26 ., Although theoretically , IT based methods such as Transfer Entropy ( TE ) and Mutual Information ( MI ) are able to detect inhibitory links , we are not aware of studies consistently reporting a successful identification of inhibitory connections ., The problem is in the incapability of distinguishing between excitatory and inhibitory links , rather than in the detection of inhibition as pointed out in the Supplementary Information ( S6 Fig ) ., Primary neurons were obtained from rat embryos ( 18 days , E18 ) from Sprague Dawley pregnant rats ( Charles River Laboratories ) ., The experimental protocol was approved by the European Animal Care Legislation ( 2010/63/EU ) , by the Italian Ministry of Health in accordance with the D . L . 116/1992 and by the guidelines of the University of Genova ( Prot . N . 24982 , October 2013 ) ., Cross-correlation ( CC ) 22 measures the frequency at which a neuron or electrode fires ( “target” ) as a function of time , relative to the firing of an event in another one ( “reference” ) ., Mathematically , the correlation function is a statistic representing the average value of the product of two random processes ( the spike trains ) ., Given a reference electrode x and a target electrode y , the correlation function reduces to a simple probability Cxy ( τ ) of observing a spike in one train y at time ( t + τ ) , given that there was a spike in a second train x at time t; τ is called the time shift or the time lag ., In this work , we use the standard definition for the cross-correlation computation , following a known normalization approach on the CC values 39 ., We define the cross-correlation as follows:, Cxy ( τ ) =1NxNy∑s=1Nxx ( ts ) y ( ts−τ ), ( 2 ), where ts indicates the timing of a spike in the x train , Nx is the total number of spikes in the x train and Ny is the total number of spikes in the y train ., Cross-correlation is limited to the interval 0 , 1 and it is symmetric Cxy ( τ ) = Cyx ( -τ ) ., The cross-correlogram is then defined as the correlation function computed over a chosen correlation window ( W , τ = -W/2 , W/2 ) ., Different shapes of cross-correlogra | Introduction, Results, Discussion, Materials and methods | Functional-effective connectivity and network topology are nowadays key issues for studying brain physiological functions and pathologies ., Inferring neuronal connectivity from electrophysiological recordings presents open challenges and unsolved problems ., In this work , we present a cross-correlation based method for reliably estimating not only excitatory but also inhibitory links , by analyzing multi-unit spike activity from large-scale neuronal networks ., The method is validated by means of realistic simulations of large-scale neuronal populations ., New results related to functional connectivity estimation and network topology identification obtained by experimental electrophysiological recordings from high-density and large-scale ( i . e . , 4096 electrodes ) microtransducer arrays coupled to in vitro neural populations are presented ., Specifically , we show that:, ( i ) functional inhibitory connections are accurately identified in in vitro cortical networks , providing that a reasonable firing rate and recording length are achieved;, ( ii ) small-world topology , with scale-free and rich-club features are reliably obtained , on condition that a minimum number of active recording sites are available ., The method and procedure can be directly extended and applied to in vivo multi-units brain activity recordings . | The balance between excitation and inhibition is fundamental for proper brain functions and for this reason is precisely regulated in adult cortices ., Impaired excitation/inhibition balance is often associated with several neurological disorders , such as epilepsy , autism and schizophrenia ., However , estimating functional inhibitory connections is not an easy task and few methods are available to identify such connections from electrophysiological data ., Here we present a cross-correlation based method to identify both excitatory and inhibitory functional connections in large-scale neuronal networks ., The method is applicable to both in vitro and in vivo spike data recordings ., Once a connectivity map ( i . e . a graph ) is obtained , we characterized the associated topology by means of classical graph theory metrics to unveil functional architecture ., In this work , we analyze in vitro cortical networks probed by means of large-scale microelectrode arrays ( i . e . , 4096 sensors ) and we derive network topologies from spike data ., The functional organization found is called “small-world and scale-free” and is the same organization found in cortical in vivo brain regions by means of different experimental methods ., We also show that to obtain reliable information about network architecture at least a network with a hundred of nodes-neurons is needed . | action potentials, medicine and health sciences, neural networks, applied mathematics, membrane potential, electrophysiology, neuroscience, simulation and modeling, algorithms, mathematics, network analysis, membrane electrophysiology, distribution curves, bioassays and physiological analysis, research and analysis methods, statistical distributions, computer and information sciences, animal cells, electrophysiological techniques, probability theory, cellular neuroscience, electrode recording, cell biology, physiology, neurons, biology and life sciences, cellular types, physical sciences, neurophysiology | null |
journal.pbio.1000621 | 2,011 | Hydrogen Peroxide Promotes Injury-Induced Peripheral Sensory Axon Regeneration in the Zebrafish Skin | Successful wound repair and regeneration requires coordination between the various cell types that make up the injured tissue ., For example , following injuries that damage both epidermis and sensory endings , wounded epidermis promotes the regeneration of nerve fibers 1 , 2 ., Conversely , complete epidermal wound healing requires the presence of sensory axons 1 , 3 ., In amphibians , innervation of the wound epidermis by nerve fibers is also essential for limb regeneration and correlates with the establishment of signaling centers 4–6 ., These observations imply that coordination between wound epidermis and sensory axons during healing and regeneration is regulated by molecular interactions between these cell types ., In mammals , peripheral axon regeneration is generally more robust than axon regeneration in the central nervous system ., Nonetheless , reinnervation in the periphery can be slow or incomplete , depending on the extent of axonal injury and on interactions with surrounding cells 7 , 8 ., Because nerve injury is often associated with damage of not only the nerve but also neighboring tissues , it has been difficult to separate autonomous and non-autonomous factors influencing axon regeneration in vivo ., Recent studies in C . elegans and zebrafish have utilized laser axotomy to precisely damage single axons in the peripheral nervous system , making it possible to assess the influence of non-neuronal tissues on axonal regeneration 9 , 10 ., Tissue damage triggers a complex cascade of signals that activate inflammatory responses and promote tissue repair 11 ., In fruit flies and zebrafish , the recruitment of immune cells to wounds is mediated by the small reactive oxygen species ( ROS ) hydrogen peroxide ( H2O2 ) , which emanates from the injury 12 , 13 ., The role of H2O2 in oxidative stress has been well studied , as high levels can have deleterious effects on the maintenance of cell homeostasis 14 ., In the nervous system , H2O2 can induce neurodegeneration through activation of pro-apoptotic pathways 15–17 ., More recently it has come to be appreciated that H2O2 can act as a signaling molecule with specific developmental and physiological functions ., H2O2 is thought to signal by oxidizing cysteine residues on target proteins , most notably phosphatases 18 , 19 ., The larval zebrafish tail fin provides an accessible setting for investigating how peripheral axon regeneration is coordinated with the healing of injured tissue and for testing whether H2O2 plays a role in these interactions ., During larval stages , zebrafish fins consist of a folded two-layered epithelium , surrounding muscle cells ( Figure S1A ) ., Zebrafish tail fins regenerate after amputation , both during larval development 20 , 21 and in adults 22 , but sensory reinnervation of regenerated fins has not been explicitly assessed ., Somatosensation at larval stages in zebrafish is accomplished by two populations of neurons: trigeminal neurons , which are located in ganglia outside the hindbrain and innervate the skin of the head , and Rohon-Beard ( RB ) neurons , which are located in the dorsal spinal cord and innervate the skin of the trunk and tail ( Figure 1A ) ., The peripheral axons of somatosensory neurons arborize between the two epithelial layers that make up the larval skin , the outer periderm and inner basal cell layers 23 ., Precisely severing a trigeminal peripheral axon after arborization is complete ( ∼36 h post-fertilization , hpf ) promotes some regenerative growth , but regenerating axons avoid their former territories and undamaged neighboring axons never sprout into these newly denervated areas 24 ., We have investigated the relationship between tissue damage and peripheral axon regeneration , using injury to the larval zebrafish tail fin as an experimental paradigm ., Amputating the fin promoted peripheral sensory axon growth , allowing the robust reinnervation of the newly regenerated fin ., This axon regeneration-promoting effect could also be elicited by ablating a few keratinocytes anywhere in the body ., H2O2 exposure mimicked the axon growth-promoting effect of keratinocyte damage , and morpholino-mediated knockdown of the H2O2-generating enzyme Duox1 inhibited axon growth-promotion by fin amputation ., Thus , H2O2 produced by damaged keratinocytes promotes the reinnervation of healing skin by sensory axons ., The caudal fins of larval zebrafish regenerate completely within a few days after amputation 20 , implying that RB peripheral axons must also regenerate to provide sensory function to the new fin ., To directly assess whether RB axons in the tail can regenerate , we imaged GFP-labeled RB arbors in the islet2b:GFP transgenic line 25 after caudal fin amputation at 3 d post-fertilization ( dpf ) ., Amputation caused immediate degeneration of axon branches near the wound ( Figure 1B , brackets ) , creating a denervated zone that regenerating axons would need to traverse to fully innervate the regenerating fin ., Despite this potential barrier , the fin was always reinnervated by RB arbors at 120 h post-amputation ( hpamp ) ( Figure 1B ) ., Three days after fin amputation , there was no detectable difference in the total amount of sensory axons in regenerated fin tips and fin tips of age-matched animals ( 6 dpf ) that were never injured , indicating that reinnervation of regenerated fins was complete ( Figure 2A ) ., Sensory reinnervation of regenerated fins was functional , since 6 dpf fish with regenerated fins responded to touch at the tip of the tail as often as uninjured control fish ( Figure 2B ) ., The observation that RB axons robustly reinnervate larval fins within a few days after amputation , despite the fact that trigeminal axon regeneration is limited after precise axotomy 24 , could be explained in either of two ways: ( 1 ) fin injury and healing promote peripheral axon growth or ( 2 ) RB neurons innervating the tail possess greater structural plasticity than trigeminal neurons ., To assess the intrinsic plasticity of RB axon arbors , we monitored axon behavior after precise laser axotomy with time-lapse imaging for 12 h ( see Figure S1B for experimental procedures ) 10 and traced the position of individual axon tips every 30 min ., Axotomy of RB neurons induced a 2-fold increase in axon activity ( axon tip displacement , including both growth and retraction ) compared to uninjured axons ( 54 . 92±2 . 72 µm , n\u200a=\u200a24 versus 32 . 47±2 . 53 µm , n\u200a=\u200a13 axon tip displacement , * p<0 . 05; compare Figures 1C and 3A; quantification in Figure 3D , Videos S2 and S1 , respectively ) , but , like trigeminal axons , regenerating RB axons avoided denervated territory ( Figure S2 ) 24 ., Notably , axon growth was balanced by retraction , so that total arbor size did not substantially increase ( Figure 3F; see Video S2 ) ., Like trigeminal axons 24 , the ability of RB axons to reinnervate former territory in the fin was improved by inhibiting Rho kinase ( unpublished data ) ., Thus , the ability of RB axons to regenerate after fin amputation is likely not due to intrinsic regenerative capacity but is probably a specific response to tissue damage ., To further investigate the influence of tissue injury on RB axon regeneration , we compared the behavior of uninjured axon arbors ( Figure 1C ) , precisely axotomized arbors ( Figure 3A ) , and injured arbors in amputated fins ( Figure 1D ) ., Fin amputation ( Video S3 ) increased total axon activity ( growth and retraction ) more than axotomy alone ( 77 . 40±4 . 03 µm , n\u200a=\u200a26 , *** p<0 . 001 ) ., Measuring the linear distance between an axon tips position just after amputation and its position 12 h later revealed that fin amputation promoted productive axon growth , since axon tips traveled farther after amputation than after precise axotomy ( 29 . 62±2 . 50 µm , n\u200a=\u200a8 , versus 8 . 42±3 . 09 µm , n\u200a=\u200a8 , ** p<0 . 01; Figure 3E ) ., Combining fin amputation with subsequent laser axotomy of a nearby RB axon branch increased the axon activity ( 83 . 74±3 . 09 µm , n\u200a=\u200a26 , ** p<0 . 01 ) and total growth ( 46 . 54±4 . 92 µm , n\u200a=\u200a13 , *** p<0 . 001 ) even further ( Figure 3B , D , E , Video S4 ) , but the amount of retraction was not dramatically altered ( Figure 3F ) ., Amputating fins significantly improved the ability of regenerating axons to innervate denervated areas ( 14 . 11±7 . 02 µm , n\u200a=\u200a8 versus 60 . 24±13 . 06 µm , n\u200a=\u200a10 , * p<0 . 05; Figure S2 ) , which is likely important for allowing regenerating arbors to traverse the denervated zone that forms just proximal to the wound after amputation ( Figure 1B , brackets ) ., Thus , fin injury increases sensory axon activity , promotes growth ( but not retraction ) , and allows axons to overcome their avoidance of denervated territories ., To determine the effective range of axon growth-promoting signals from injured tissue , we axotomized axons distant ( >50 µm ) from the amputation site ( Figure 3C , Video S5 ) ., These axons did not grow significantly better than precisely severed axons in uninjured tissue , since neither axon activity nor linear growth distances were increased by distant amputation ( axon activity: 38 . 67±2 . 85 µm , n\u200a=\u200a10 , p\u200a=\u200ans>0 . 05; linear distance\u200a=\u200a15 . 63±4 . 42 µm , n\u200a=\u200a9; p\u200a=\u200ans>0 . 05; Figure 3D , E ) ., Thus , growth-promoting signals emanating from injured tissue likely function at short range ., To define the time window during which axons can respond to regeneration-promoting signals , we axotomized RB arbors at different time points after amputation ., Axon activity was most enhanced when arbors were axotomized at 3 h post-amputation ( 114 . 6±7 . 04 µm , n\u200a=\u200a10 , ** p<0 . 01 ) , but axotomy at 6 h post-amputation did not increase axon activity ( 61 . 20±6 . 45 µm , n\u200a=\u200a10 , p\u200a=\u200ans>0 . 05; Figure S3A ) , as compared to axotomy alone ., This observation suggests that axon growth-promoting signals are transiently emitted from the wound , rather than continuously from regenerating fin tissue ., To assess whether the size of the severed arbor fragment influenced the amount of axon activity induced by amputation , we traced degenerated fragments in three dimensions to measure their total length and plotted length as a function of axon activity ., Size of the axotomized arbor did not correlate with axon activity ( Figure S3B ) ., To identify the origin of axon growth-promoting signals , we ablated individual muscle cells or keratinocytes in the fin of larvae expressing cell type-specific reporter transgenes that highlight each tissue 26 , 27 ., Ablating muscle cells did not promote axon growth ( 25 . 72±3 . 65 µm , n\u200a=\u200a10; Figure 4A , E ) , but ablating ≥3 keratinocytes prior to axotomy provoked robust axon regeneration in both the fin ( 70 . 75±6 . 14 µm , n\u200a=\u200a11 , Figure 4B , E ) and head ( 73 . 81±20 . 95 µm , n\u200a=\u200a4; Figure 4C , D , E ) ., However , ablating a single keratinocyte in either the fin ( 44 . 34±2 . 35 µm , n\u200a=\u200a10 , *** p<0 . 001 ) or the head ( 27 . 62±1 . 94 µm , n\u200a=\u200a14 , ** p<0 . 01 ) did not promote axon regeneration ., This result suggests that a threshold of injury-induced signals is required to promote growth and reinnervation by RB and trigeminal axons ., The recently reported observation that zebrafish larval fin amputation produces high levels of hydrogen peroxide ( H2O2 ) at the wound margin 13 prompted us to investigate whether H2O2 contributes to the promotion of axon regeneration by keratinocyte injury ., By monitoring H2O2 with a chemical sensor ( pentafluorobenzenesulfonyl fluorescein ) , we first verified that , like fin amputation ( Figure 5A ) , laser ablating several keratinocytes produced detectable levels of H2O2 around the wound ( Figure 5B ) ., Ablation of 1–2 keratinocytes did not produce levels of H2O2 sufficient to detect with the chemical sensor , but ablating ≥3 keratinocytes generated detectable levels of H2O2 at the wound margin ( Figure 5C ) , indicating that the severity of the injury correlates with the amount of H2O2 produced ., To test whether H2O2 can promote axon regeneration , we added 3 mM H2O2 ( 0 . 01% ) to the larval media ( the highest concentration of H2O2 at which most embryos survived and developed normally , see Figure S4 for survival rates ) ( Figure 6A , Video S6 ) ., The addition of H2O2 to uninjured larvae significantly promoted some axon activity ( untreated , uninjured: 32 . 47±2 . 53 µm versus H2O2 uninjured: 72 . 30±1 . 94 µm , *** p<0 . 001; Figure 6D ) ., Adding H2O2 for 3 or 12 h to larvae in which RB axon arbors had been axotomized increased axon activity variably but significantly , compared to axotomy in untreated animals ( 3 h H2O2: 122 . 1±8 . 81 µm , n\u200a=\u200a6; 12 h H2O2: 101 . 4±3 . 09 µm , n\u200a=\u200a10 , versus untreated 54 . 92±2 . 72 µm , n\u200a=\u200a24 , ** p<0 . 01 each; Figure 6D ) ., The linear growth distances of axotomized arbors were also increased by H2O2 ( 3 h: 43 . 58±6 . 06 µm , n\u200a=\u200a5; 12 h: 30 . 04±2 . 25 µm , n\u200a=\u200a8 , versus untreated: 5 . 46±3 . 78 µm , n\u200a=\u200a5 , ** p<0 . 01 each; Figure 6E ) ., Thus , H2O2 is sufficient to promote axon regeneration and does not need to be present in a gradient for this effect , as has been proposed for its role in leukocyte recruitment 13 ., To test whether H2O2 is required for injury-induced axon growth , we blocked H2O2 production and monitored axon regeneration after fin amputation by downregulating the primary enzyme generating H2O2 in the larval fin , Dual oxidase 1 ( Duox1 ) , using a previously characterized morpholino ( duox1-MO ) ( Figures 5A , S5A ) 13 ., Injecting this morpholino into embryos prevented the promotion of axon activity by fin amputation ( 23 . 89±3 . 29 µm , n\u200a=\u200a11 , ** p<0 . 01 ) ( Figure 6B , F , Video S7 ) ., Interestingly , fin regeneration was also compromised in duox1 morphants , potentially reflecting a role for axon innervation in fin regeneration , similar to limb regeneration in amphibians 4 ., Treating amputated morphant larvae with 1 . 5 mM H2O2 for 12 h rescued the deficit in axon reinnervation observed in the morphants ( 102 . 3±5 . 6 µm , n\u200a=\u200a5 , ** p<0 . 01 ) ( Figure 6C , F , Video S8 ) ., Due to the toxicity of prolonged H2O2 treatment , we unfortunately could not assess whether such rescued morphants also regenerated their fins ., Blocking H2O2 production with the duox1-MO did not affect growth and retraction induced by axotomy alone ( Figure 6D ) , suggesting that cell-intrinsic mechanisms through which axotomy induces axon activity may be regulated by different pathways ., To minimize the possibility that duox1-MO toxicity inhibited axon growth following axotomy , we repeated this experiment with co-injection of a morpholino targeting p53 , which inhibits apoptosis 28 , as was done in a previous study with the duox1-MO 13 ., Like in larvae injected with duox1-MO alone , axon growth promotion by amputation was blocked in larvae injected with both p53-MO and duox1-MO , compared to larvae injected with p53-MO alone ( Figures 6F , S5B–D ) , supporting the idea that the duox1-MOs effect on regeneration is not due to cellular toxicity ., Together these results indicate that Duox1-mediated H2O2 production is necessary for the promotion of injury-induced axon growth ., To determine where Duox1 is required to promote axon regeneration , we created genetic chimeras by transplanting cells at the blastula stage from donor embryos injected with duox1-MO into uninjected host embryos ( Figure 7A ) ., Donor embryos were transgenic for a somatosensory GFP reporter ( sensory:GFP ) and host embryos were transgenic for a keratinocyte RFP reporter ( Krt4:RFP; previously termed Krt8 ) 27 , 29 ., At larval stages , we ablated wildtype RFP-labeled keratinocytes in these chimeras and axotomized nearby duox1 morphant peripheral sensory arbors ( n\u200a=\u200a7 RB neurons , Figure 7B , Video S9 ) ., Keratinocyte ablation in these animals significantly promoted axon regeneration , demonstrating that Duox1 is required non-autonomously to achieve the full level of axon growth promotion by keratinocyte ablation ( Figure 7C ) ., This result also verified that the ability of the duox1-MO to block regeneration was not due to morpholino toxicity in the neuron ., The promotion of axon growth by H2O2 could in principle result from the direct activation of axon growth or the repression of axon growth inhibitors , such as those that arise after initial growth stages to stabilize axonal structure 24 ., To address this issue , we examined regeneration at 30 hpf , when axons are developing and repellants are presumably not present ( Figure S6 ) ., Axotomy at this stage increased axon activity ( ** p<0 . 01 ) , but the addition of 3 mM H2O2 to developing 30 hpf larvae did not further increase activity when compared to untreated larvae ., Conversely , knockdown of duox1 did not significantly change axon activity after fin amputation in 30 hpf larvae ( WT versus duox1-MO , p\u200a=\u200ans>0 . 05 ) ., These results support the notion that H2O2 acts by blocking axon growth inhibition , since it only influences regeneration at stages when inhibitors are present ., Interestingly , a study in chick showed that axon growth promotion by skin wounds was also only effective at late developmental stages 1 ., H2O2 promotes immune cell recruitment to wounds in developing fruit fly embryos 12 and during early inflammatory responses to fin amputation in larval zebrafish 13 ., To test whether inflammation and axon growth are linked or independent effects of H2O2 signaling , we assessed axon growth in homozygous cloche mutants , which lack blood cells 30 , and macrophage recruitment in larvae injected with ngn1-MO , which lack somatosensory neurons 31 ., Amputation promoted axon growth in the absence of blood ( 91 . 16±10 . 44 µm , n\u200a=\u200a9; Figure 6F , Video S10 ) and macrophages homed to the wound in the absence of sensory neurons ( 1 . 5±0 . 42 macrophages expressing lysC:GFP at the wound margin within 1 h of amputation , n\u200a=\u200a6 ) , similar to wildtype ( 1 . 0±0 . 43 macrophages at the wound margin within 1 h of amputation , n\u200a=\u200a7 , p\u200a=\u200ans>0 . 05; Figure S7 ) , indicating that these two processes are independent of each other ., Our results demonstrate that skin injury promotes the growth of axons near the wound , an effect that is mediated by H2O2 ( Figure 8 ) ., Keratinocyte ablation and genetic chimera experiments suggested that the axon growth-promoting effects of H2O2 require its production in keratinocytes ., Similarly , in axolotl and chick , wound epidermis attracts axons 1 , 2 and damage to human skin co-cultured with rat dorsal root ganglia promotes regeneration of axons at the dermal/epidermal interface 32 ., It will be interesting to determine whether H2O2 also plays a role in these phenomena ., Intriguingly , H2O2 improves hippocampal neurite outgrowth in culture 33 ., In C . elegans , a mutation in pxn-2 , which encodes an extracellular peroxidase , improves regeneration of mechanosensory axons 34 ., In zebrafish , H2O2 may be signaling directly to axons , altering the extracellular matrix , or eliciting a second signal from keratinocytes to promote axon growth , but does not require leukocytes ( Figure 8 ) ., Assessing whether application of H2O2 to somatosensory neurons in culture can improve axon growth , as has been reported for hippocampal neurons in culture 33 , could help resolve whether H2O2 acts directly or indirectly on axons to influence their regeneration ., In summary , we have found that wounded epidermis promotes somatosensory axon regeneration in zebrafish larvae and that H2O2 is a critical mediator of this effect ., Since this effect does not require the presence of leukocytes , we propose that H2O2 plays two independent roles during wound healing: promoting axon growth and mediating leukocyte recruitment ., Thus , one signaling molecule emitted from injured tissue helps coordinate wound healing with functional recovery of skin ., Zebrafish embryos were obtained from Nacre 35 , AB ( wildtype ) , Line Mü4435_64 26 , cloche ( clom39 ) 30 , lysC:GFP 36 , sensory:GFP 29 , and islet2b:GFP 25 fish ., Embryos and larvae were treated with 0 . 15 mM Phenylthiourea ( PTU ) to prevent pigment formation ., TEM was performed according to Rieger & Köster , CSH Protocols Vol ., 2 ( doi:10 . 1101/pdb . prot4772 , 2007 ) ., Zebrafish larvae were anesthetized in 0 . 01% Tricaine and mounted in a sealed chamber in 1 . 2% low-melting agarose ( Sigma , St . Louis , MO ) ., Details of the mounting and imaging techniques are described elsewhere 10 ., Larvae were imaged for 12 h using a 20× air objective ., Stacks were scanned every 30 min in 3 µm intervals ., Imaging was performed with 6–10 larvae per session on an LSM 510 confocal microscope ( Zeiss ) with an automated stage and Multitime software ., Larvae were maintained at 28 . 5°C using a stage heater ., Maximum intensity projections of confocal stacks were compiled using Zeiss software and further processed using Adobe Photoshop , NIH open source software Image J 1 . 34S ( Abramoff , NIH Open Source software ImageJ , 2004 ) , and Quick Time Player 7 Pro ., For time-lapse imaging of peripheral sensory axon regeneration in H2O2 solution , 0 . 005%–0 . 01% H2O2 ( 1 . 5–3 mM ) was added to the larval media 1 h prior to axotomy ., Larvae were maintained in H2O2 solution for 3 or 12 h of time-lapse imaging ., See also Figure S1B for timeline of experiments ., All transgenes were constructed using the Gateway ( Invitrogen ) tol2kit created by the lab of Chi-Bin Chien 37 ., Tol2CREST3-Gal4VP16-14xUAS-EGFP: The somatosensory neuron-specific CREST3 enhancer 38 was cloned into the 5′ Gateway vector ( p5E ) , Gal4VP16-14xUAS 39 into the middle vector ( pME ) , and EGFP-SV40pA into the 3′ vector ( p3E ) ., Elements were recombined together with the Tol2 destination vector ( pDESTTol2 ) ., Tol2CREST3-LexA-LexAop-EGFP: LexAVP16-SV40pA and four copies of the LexAop 40 were cloned into the middle Gateway vector ( pME ) and recombined with p5E-CREST3 and p3E-EGFP-SV40pA to generate Tol2CREST3-LexA-4xLexAop-EGFP ., Approximately 15 pg of CREST3-Gal4VP16-14xUAS-EGFP or CREST3-LexA-LexAop-EGFP plasmids were co-injected with ∼240 pg of Tol2 41 transposase mRNA into 1-cell stage embryos of wildtype AB or Nacre strains or into the Gal4-UAS muscle reporter line Tg ( Mü4435_64 ) 26 , respectively ., A similar amount of CREST3-Gal4VP16-14xUAS-EGFP was co-injected with 10 pg of Krt4:RFP and Tol2 transposase mRNA for keratinocyte ablations ., To knock down expression of p53 28 , duox1 13 , and ngn1 31 , 50 nM of each modified antisense oligonucleotide was injected into 1-cell stage embryos ., Knockdown of duox1 by morpholino injection was verified with RT-PCR , using published primers 13 ., Ten larvae at 3 dpf were pooled for RNA isolation and subsequent RT-PCR ( see also Figure S5A ) ., To determine the sublethal concentration of H2O2 ( Fisher Biotech , 30% in water ) to use in larval experiments , we identified the maximum concentration at which 100% of larvae were viable for a minimum of 12 h ., Groups of five larvae were incubated in serial dilutions of H2O2 from 0 . 003% to 30% and viability was assessed 12 h later ., The EC50 was determined to be ∼0 . 03% ( Figure S4 ) ., Larvae survived without any morphological abnormalities at 3 mM H2O2 ( 0 . 01% ) or less ., For rescue experiments , we used a lower concentration of H2O2 ( 1 . 5 mM ) to maintain optimal viability ., To detect the presence of H2O2 after amputation or ablation , 5 µM of the H2O2 sensor pentafluorobenzenesulfonyl fluorescein ( Santa Cruz Biotechnology ) was added 1 h prior to injury ., Larvae were exposed to the sensor throughout the imaging procedure up to 12 h ., Fluorescence was detected at 488/505 nm ., To create chimeras between wildtype and duox1-morphants , a few blastula cells ( 1 , 000–cell stage ) were transplanted from sensory:GFP transgenic embryos injected with duox1 morpholino into Krt4:RFP wildtype transgenic embryos ., Larvae were screened for sensory-specific GFP expression ( Duox1-negative neurons ) and red fluorescence in keratinocytes ( H2O2-positive skin ) ., Axons were axotomized and imaged as described above ., Macrophages were imaged for 12 h in lysC:GFP 36/islet2b:GFP 25 double-transgenic zebrafish larvae ( 78 hpf ) , which were either wildtype or injected with 50nM of ngn1 morpholino 31 to inhibit sensory neuron development ., New macrophages that arrived within 1 h after amputation at the amputation margin were counted and compared between both groups , similar to 13 ., Axon activity was measured by tracing the movements of the 10 axon tips that grew most over a 12 h time window using Image J 1 . 34S and the Image J Manual Tracking software plugin ( F . Cordelires , Institut Curie , Orsay , France ) ., Projected images were adjusted for movement of the specimen , using the Image J StackReg plugin ( P . Thévenaz , Swiss Federal Institute of Technology , Lausanne , Switzerland ) ., Measurements were made from projections of 24 time points recorded at 30 min intervals , assuming that axon tips move in a two-dimensional plane ., A minimum of 10 axon tips per arbor and specimen were traced ., The linear distances of axon growth were quantified using the Zeiss LSM 510 software and ImageJ analysis tool by measuring the distance between the growth cone position in the first ( 1 h ) and last ( 12 h ) stack ., To quantify reinnervation of denervated territory , NeuroLucida software ( Microbrightfield , Williston , VT ) was used to generate tracings of individual Rohon-Beard axons in the fin skin from confocal stacks at 30 min ( first recorded time point ) and 12 h ( last recorded time point ) ., These tracings were overlaid and length measurements were used to quantify the percentage of the axon that entered denervated territory ., To minimize distortion caused by developmental growth , images were aligned at the closest shared branch point proximal to the site of axotomy ., Statistical analyses were performed using Prism 4 ( GraphPad Software Inc . ) ., Unpaired , two-tailed Students t-tests were used for comparisons of two groups ( Figures 1E and S2 ) ., One-way ANOVA and Dunnetts ( comparing groups to a control group ) or Bonferronis ( comparing groups to one another ) post-tests were performed as indicated in each figure ., Significance was set to p<0 . 05 ., All graphs show the standard error of the mean ., Confocal images were loaded into ImageJ software and converted to 8-bit images ., A binary image was created and the mean pixel values in a 50×50 µm field in the distal fin portion measured to determine the axon density ., Images were exported as tiff files from the LSM software ( Zeiss ) and loaded into the ImageJ software ., Axon tips were traced as described above and individual movements were designated as growth or retraction within each 30 min interval ., The total length of growth and retraction for each arbor was calculated for a 12-h period and a mean value of all traced axon tips derived ( n\u200a=\u200a4 axon tips/4 axons\u200a=\u200a16 tracings total ) ., The detached distal portions of axotomized axons were traced using Neurolucida software ( MBF Bioscience ) to determine the total combined length of all the branches in the detached arbor ., The length was plotted against axon activity of the parent axon during the regeneration phase ( 12 h ) ., Larvae were placed in a petri dish and tapped with an insect pin at the distal tip of the caudal fin and escapes were recorded ., Two groups were compared: wildtype uninjured larvae at 6 dpf and age-matched wildtype larvae whose fins were amputated 3 dpf . | Introduction, Results/Discussion, Materials and Methods | Functional recovery from cutaneous injury requires not only the healing and regeneration of skin cells but also reinnervation of the skin by somatosensory peripheral axon endings ., To investigate how sensory axon regeneration and wound healing are coordinated , we amputated the caudal fins of zebrafish larvae and imaged somatosensory axon behavior ., Fin amputation strongly promoted the regeneration of nearby sensory axons , an effect that could be mimicked by ablating a few keratinocytes anywhere in the body ., Since injury produces the reactive oxygen species hydrogen peroxide ( H2O2 ) near wounds , we tested whether H2O2 influences cutaneous axon regeneration ., Exposure of zebrafish larvae to sublethal levels of exogenous H2O2 promoted growth of severed axons in the absence of keratinocyte injury , and inhibiting H2O2 production blocked the axon growth-promoting effects of fin amputation and keratinocyte ablation ., Thus , H2O2 signaling helps coordinate wound healing with peripheral sensory axon reinnervation of the skin . | Touch-sensing neurons project axonal processes that branch extensively within the outer layers of skin to detect touch stimuli ., Recovering from skin injuries thus requires not only repair of damaged skin tissue but also regeneration of the sensory axons innervating it ., To study whether skin wound healing is coordinated with sensory innervation , we compared the regeneration of severed sensory axons innervating larval zebrafish tail fins with and without concomitant injury to surrounding skin cells ., Severed axons regenerated more robustly when nearby skin cells were also damaged , suggesting that wounded skin releases a short-range factor that promotes axon growth ., The reactive oxygen species hydrogen peroxide ( H2O2 ) is known to be produced by injured cells , making it a candidate for mediating this signal ., We found that adding exogenous H2O2 improved the regeneration of severed axons ., Conversely , blocking H2O2 production prevented the axon growth-promoting effect of skin injury ., Thus , H2O2 promotes axon growth after skin damage , helping to ensure that healing skin is properly innervated . | cell biology/neuronal and glial cell biology, neuroscience/sensory systems, cell biology/cell signaling, cell biology/morphogenesis and cell biology, cell biology/developmental molecular mechanisms, cell biology/neuronal signaling mechanisms, neuroscience/neurodevelopment, developmental biology/neurodevelopment, neuroscience/neuronal signaling mechanisms, neuroscience/neurobiology of disease and regeneration | Production of H2O2 by injured zebrafish skin cells promotes the regeneration of nearby somatosensory axon terminals, thus coordinating wound healing of the skin with sensory reinnervation. |
journal.pntd.0002303 | 2,013 | Assessment of Transmission in Trachoma Programs over Time Suggests No Short-Term Loss of Immunity | The World Health Organization has targeted trachoma for elimination by the year 2020 1 ., Repeated mass oral azithromycin distributions have been a cornerstone of the treatment strategy ., Theoretically , repeated treatments may eventually eliminate infection from even the most severely affected areas 2 , 3 ., In practice , distributions have dramatically reduced the prevalence of infection in a number of locations 3 , 4 , 5 , 6 , 7 , 8 , 9 ., However , there remains concern that resistance may develop or that loss of immunity may prevent complete elimination 10 , 11 , 12 , 13 , 14 ., While no stable chlamydial drug resistance has yet been observed , a loss of immunity is possible ., Individuals who had not been exposed to infection recently might have less protection than they had had when the infection was more prevalent ., An increase in transmission during the course of a program could indicate loss of immunity ., Multiple communities would need to be monitored , to assess whether random fluctuations may explain observed differences over time ., Here , we analyzed multiple communities from the Tanzanian portion of the Program for the Rapid Elimination of Trachoma trial ( PRET 15 ) using a stochastic model of transmission , to assess the initial reproductive number for transmission over time ., Communities were monitored as part of a cluster-randomized , trachoma treatment trial in Tanzania 15 , 16 ., In brief , 32 communities in Tanzania were randomized in a two by two factorial design ., The first factor was the use of standard versus enhanced coverage with annual mass antibiotic treatment; the second factor was the use or disuse of a rule whereby mass antibiotic administration would be discontinued based on ongoing monitoring ., In fact , the use of this rule never led to discontinuation of mass antibiotic administration during the first three years ., Thus , all 32 communities received treatment at baseline , 12 , and 24 months ., A baseline census was conducted in all 32 communities , and again at 12 , 24 , and 36 months ., One hundred randomly selected children aged 0–5 years were examined at baseline , and at 6 , 12 , 18 , 24 , 30 , and 36 months post baseline ., A dacron swab was passed 3 times over their inverted right upper conjunctiva , and processed for the presence of chlamydial DNA as previously described 15 ., Our stochastic transmission model was fit to the estimated prevalence of infection at 6 , 12 , 18 , 24 , 30 , and 36 months ., The study received ethical approval from institutional review board ( IRB ) of the Johns Hopkins University School of Medicine , the University of California San Francisco , and the Tanzanian National Institute for Medical Research , and was carried out in accordance with the Declaration of Helsinki ., All subjects provided informed consent ., The informed consent given was oral , because, 1 ) verbal consent is the most ethical way to obtain consent , due to the high illiteracy rates in the study area ,, 2 ) IRB approved the use of the oral consent procedure for this study ,, 3 ) this oral consent is documented on the registration form for each study participant prior to examination in the field ., We constructed a stochastic transmission model of transmission of Chlamydia trachomatis infection over time 17 , 18 , 19 , 20 ., For community j ( j\u200a=\u200a1 , … , 32 ) , we assumed a population of size Nj , taken from the number of pre-school children found in the census at the time of treatment ( baseline , 12 months , or 24 months ) ., We assumed a classical SIS ( susceptible-infective-susceptible ) model structure , assuming that the force of infection is proportional to the prevalence of infection in the population with proportionality constant β , and a constant per-capita recovery rate γ ( month−1 ) ., Between periods of treatment , we assumed that the probability pi ( t ) that there are i infectives in the population obeyed the following equations: ( 1 ) andThese equations were applied to each village j , though we have suppressed a subscript j for clarity in Equation ( 1 ) ., To estimate the transmission coefficient , we used data collected six months and twelve months after each treatment ., The model was fit to each of three years ., For comparing transmission rates , we initialized the model with observations taken six months after treatment , and we estimated the transmission parameter based on values observed six months after that ., Thus , we modeled the time periods from 6 to 12 months , 18 to 24 months , and 30 to 36 months ., For each year , initial values for pi ( t ) were determined from as follows ., From a population of size Nj of which the number Y of infectives equals i , the probability that s positives are observed from a sample of size Mj are sampled is given by the hypergeometric distribution: ., We assumed a beta-binomial prior ( where the shape parameters and were computed from the observed distribution of infection of 32 villages at 6 months , 18 months and 30 months , and B ( x , y ) is the beta function 21 ) ) for all values of y ., Application of Bayes theorem yields ( 2 ) For each community j , we used the most recent census data to determine the community size Nj ., The initial condition was determined from Equation ( 2 ) , and the system numerically integrated for six months ., Let Sj be the number of positive individuals detected in the sample at the end of the period ( for community j ) ., Given the number i of infected individuals , we computed the probability of the observed data according to ( where Mj here denotes the sample size at the end of the period ) ., We assumed independent communities , and thus we might maximize the sum of the logarithm of the above expressions ( summing over all communities ) ., We assumed specific values of γ ( see Table, 1 ) and estimated the value of β that maximizes the total loglikelihood given ., Standard errors were obtained from the observed Fisher information ., We estimated the change , m ( year−1 ) , in the transmission coefficient per year by finding maximum likelihood estimates of β1 and m ( with β2\u200a=\u200aβ1+m , β3\u200a=\u200aβ1+2m ) ., For statistical comparison , we also used all time periods to estimate a single ( constant ) transmission coefficient for all three periods ., The basic reproduction number is given by =\u200aβ/γ; thus , estimated values of may be computed by dividing the estimated transmission coefficient by γ ., The estimated annual change in the basic reproduction number , , may be computed by ., We also estimated an alternative model in which instead of varying the transmission coefficient over time , we instead assumed a constant transmission coefficient , and instead modeled the recovery rate in year i ( i\u200a=\u200a1 , 2 , 3 ) according to , where ( month−1 year−1 ) is the annual change in recovery rate ., Previous models have estimated the duration of infection 1/γ to be from 3 months to 12 months 17 , 19 , 20 ., As a base case , we assumed a mean duration 1/γ of 6 months; as a sensitivity analysis , we varied the mean duration from 3 to 18 months ., All calculations were performed using R ( version 2 . 14 . 1 , R Foundation for Statistical Computing , Vienna , Austria ) ., The numbers of 0–5 year-old children tested for the presence of ocular chlamydia were 3199 ( baseline ) , 3198 ( month 6 ) , 3191 ( month 12 ) , 3200 ( month 18 ) , 3199 ( month 24 ) , 3194 ( month 30 ) , and 3153 ( month 36 ) ., The estimated prevalence of ocular chlamydial infection by PCR at baseline was 22 . 0% ( standard deviation 10 . 1% ) , at 6 months 10 . 5% ( SD 4 . 7% ) , at 12 months 13 . 0% ( SD 6 . 4% ) , at 18 months 7 . 1% ( SD 4 . 4% ) , at 24 months 8 . 6% ( SD 7 . 3% ) , at 30 months 3 . 5% ( SD 2 . 5% ) , and at 36 months 4 . 7% ( SD 3 . 3% ) ., Assuming a mean duration of infection of six months with beta-binomial prior ( choosing the shape parameters to match the observed mean and variance of all villages cross sectionally at each initialization time ) , we found the basic reproduction number to be =\u200a1 . 39 ( 95% CI: 1 . 28 to 1 . 49 ) ., For the first year , =\u200a1 . 40 ( 95% CI: 1 . 26 to 1 . 55 ) ; for the second year , =\u200a1 . 38 ( 95% CI: 1 . 09 to 1 . 67 ) , and for the third year , =\u200a1 . 35 ( 95% CI: 0 . 92 to 1 . 78 ) ., The estimated change per year in the reproductive number , , was found to be −0 . 025 , 95% CI −0 . 167 to 0 . 117 , see Table 1 ., Similar findings were obtained when we assumed other values for the mean duration of infection: three months , =\u200a−0 . 041 ( 95% CI: −0 . 122 to 0 . 039 ) year−1; twelve months , =\u200a0 . 012 ( 95% CI: −0 . 248 to 0 . 273 ) year−1; eighteen months , =\u200a0 . 054 ( 95% CI: −0 . 326 to 0 . 434 ) year−1 ., Regardless of the assumed duration of infection , we find point estimates for the annual change in the basic reproduction number which are near zero ., The confidence intervals are wider when a longer duration of infection is assumed , and these intervals include zero ( i . e . , no change ) in every scenario ., Similar findings were obtained when a different choice of prior was used ., Specifically , we assumed a uniform distribution as the prior distribution for the number of infected individuals; this yielded an overall =\u200a1 . 30 ( 95% CI: 1 . 19 to 1 . 41 ) based on a pooled estimate assuming a constant β for all three years ( i . e . assuming m\u200a=\u200a0 ) and a mean duration of infection of six months ., The corresponding estimate for the estimated change per year in the reproductive number is =\u200a−0 . 084 ( 95% CI: −0 . 227 to 0 . 058 ) year−1 ., Choosing other values of the mean duration together with the uniform prior similarly yielded the following results: three months , =\u200a−0 . 072 ( 95% CI: −0 . 152 to 0 . 009 ) year−1; twelve months , =\u200a−0 . 103 ( 95% CI: −0 . 366 to 0 . 161 ) year−1; eighteen months , =\u200a−0 . 118 ( 95% CI: −0 . 502 to 0 . 266 ) year−1 ., We estimated the change in the recovery rate γ , assuming a constant transmission coefficient ( the optimal value when infection duration was assumed to be 6 months ) ., The estimated linear trend in recovery rate was found to be 0 . 013 ( 95% CI: −0 . 009 to 0 . 035 ) month−1 year−1 , with the estimated recovery rate in the first year given by 0 . 177 ( 95% CI: 0 . 154 to 0 . 20 ) month−1 year−1 ., This model yields a substantially similar interpretation as the previous model ., Using a transmission model and data collected from a 32-community , cluster-randomized clinical trial in Tanzania , we found no evidence of increased transmission from the 1st through the 3rd year of treatment ., In fact , our estimates of the reproduction number of the infection were very similar for each year , suggesting no loss of immunity ., Others have proposed an arrested immunity hypothesis , in which the development of protective immune responses is decreased as the duration of chlamydial infection is decreased ., It has been suggested that an increased incidence of infection in the presence of a decreased seroprevalence in Finland and British Columbia is due to this phenomenon 10 , 22 ., Trachoma programs have offered an ideal setting to test this hypothesis ., A study in Vietnam suggested that a single community with more antibiotic treatment had more rapid return of infection than another less treated community , and that this may be due to a loss of immunity 11 ., Without a larger number of treated communities , it was not possible to assess whether the magnitude of this paradoxical result would have been expected by chance alone 23 ., In this present study , the large number of longitudinally monitored communities offered more power , yet we were unable to document any evidence of increased transmission with more treatment ., There are several reasons this analysis of these data might fail to detect an increase in transmission , even if such an increase in fact exists ., Two years may not be long enough for immunity to wane , although the period in which an earlier study suggested waning was only one year 11 ., Uncertainty in the average duration of infection is a potential source of model misspecification , although a sensitivity analysis suggests that the reproductive number and estimated change are not sensitive to the value of infection duration ., Even though the trial from which these data came is one of the larger trachoma studies performed , 32 communities may not be a large enough sample size to detect a modest increase in transmission ., It is possible that a loss of immunity did occur , but that any effect on transmission was balanced by a decrease in transmission due to other factors; other studies have reported that per-infectious case transmission may decrease with decreasing prevalence , perhaps due to a decrease in the diversity of strains at lower prevalence 19 , 24 ., Finally , we have assumed that transmission is proportional to the number of infectious cases and number of susceptible cases ( mass action ) ; if this is not the case , then this may have masked increased transmission at the later , lower prevalence 19 ., Models have predicted that if transmission per infectious case remains constant , repeated distributions can eliminate infection from even the most severely affected communities 2 , 3 ., Longitudinal studies have confirmed that local elimination is possible 5 , 6 , 25 , 26 , 27 ., However , these successes might not hold in the future , if antibiotic resistance were to develop , or if a loss of immunity resulted in increased transmission ., The absence of a short term increase in transmission as the prevalence decreases is good news for trachoma programs . | Introduction, Methods, Results, Discussion | Trachoma programs have dramatically reduced the prevalence of the ocular chlamydia that cause the disease ., Some have hypothesized that immunity to the infection may be reduced because of program success in reducing the incidence of infection , and transmission may then increase ., Longitudinal studies of multiple communities would be necessary to test this hypothesis ., Here , we quantify transmission using an estimated basic reproduction number based on 32 communities during the first , second , and third years of an antibiotic treatment program ., We found that there is little to no increase in the basic reproduction number over time ., The estimated linear trend in the basic reproduction number , , was found to be −0 . 025 per year , 95% CI −0 . 167 to 0 . 117 per year ., We are unable to find evidence supporting any loss of immunity over the course of a 3-year program ., This is encouraging , as it allows the possibility that repeated mass antibiotic distributions may eliminate infection from even the most severely affected areas . | Trachoma , caused by repeated infections by the ocular strains of Chlamydia trachomatis , is the most common infectious cause of blindness in the world ., Treatment for trachoma includes mass azithromycin treatments to the entire community ., To reduce the prevalence of infection , the World Health Organization ( WHO ) advocates at least three annual community-wide distributions of oral antibiotics in affected areas , with further mass treatments based on the prevalence of trachoma ., Trachoma programs have dramatically reduced the community prevalence of infection , and some have argued that lowered prevalence of infection may lead to reductions in immunity , and that less immunity may in turn lead to increased transmission from what infection remains ., Here , we used a stochastic transmission model to analyze data collected from a 3-year antibiotic treatment program ( a 32-community , cluster-randomized clinical trial in Tanzania ) to assess whether or not transmission actually increases during elimination campaigns ., We found no evidence supporting any increase in transmission over the course of the program ., The absence of a short term increase in transmission as the prevalence decreases is good news for trachoma programs . | complex systems, medicine, infectious diseases, trachoma, mathematics, population modeling, ophthalmology, applied mathematics, neglected tropical diseases, infectious disease modeling, biology, computational biology | null |
journal.pntd.0006379 | 2,018 | Dynamics of cholera epidemics from Benin to Mauritania | Seven cholera pandemics have been documented since 1817 1 ., The disease has plagued every continent , spreading along trade routes via both land and sea 1 ., Current epidemics are however localized to South Asia , Haiti , and Sub-Saharan Africa 2 ., Since the ongoing pandemic first reached West Africa in 1970 1 , outbreaks have been repeatedly reported throughout the region 2 ., However , only a few small-scale studies have investigated the dynamics of recent cholera epidemics in West Africa ., Overall , cholera outbreaks throughout the study region have displayed markedly diverse patterns depending on the country ., Certain countries such as Benin and Togo have reported cholera cases nearly every year albeit with relatively low incidence 3 , 4 ., By contrast , many other countries to the northwest such as Gambia , Senegal , and Mauritania have experienced marked multi-year lulls 5–7 ., Many large epidemics erupted on the heels of violent civil conflicts that engendered humanitarian and public health crisis or massive population movement such as that in Senegal in 2004–2006 7 ., The large majority of cases were also reported from large cities following increased rainfall 7–10 ., Although many studies were limited to a single outbreak/neighborhood or a short time period , common risk factors were found across the region: crowded living conditions , poor sanitation , and limited access to potable drinking water 3 , 8–11 ., Cholera is contracted by consuming food or water contaminated with toxigenic Vibrio cholerae O1 or the derivative Vibrio cholerae O139 12 ., Numerous V . cholerae non-O1 and some O1 serogroups lacking the cholera toxin are autochthonous in seawaters worldwide 13 ., V . cholerae non-O1 serogroups have been found associated with a variety of aquatic flora and fauna , notably copepods 14 ., In the Bay of Bengal , elevated seawater temperatures , copepod and plankton blooms , and rainfall have been shown to correlate with increased concentrations of V . cholerae in the environment 15–17 ., In West Africa , a study has addressed the relationship between climate inter-annual variability and cholera in Nigeria , Benin , Togo , Ghana , and Ivory Coast over a 20-year period ., From 1987–1994 , they observed temporospatial synchrony between cholera incidence and rainfall in all countries except Ivory Coast 18 ., Cholera has thus been depicted as a waterborne disease driven by ecological factors 12 , 19 ., However , despite technological improvements , a perennial aquatic reservoir of cholera-causing V . cholerae O1 has yet to be identified in West Africa 20 ., We applied an integrated approach to describe the dynamics of cholera epidemics , including the identification of hotspots , and investigate factors that may influence the disease in several coastal West African countries from Benin to Mauritania ., We analyzed weekly cholera outbreak evolution ( at the district or commune level ) throughout the region between 2009 and 2015 and field investigations in Benin , Togo , Ghana , Ivory Coast , Sierra Leone , and Guinea ., We performed molecular epidemiology analysis ( MLVA ( Multi-Locus VNTR Variable Number Tandem Repeat Analysis ) and whole-genome sequencing ) of V . cholerae isolates from the majority of outbreaks affecting the study region since 2010 ., Such molecular epidemiology analysis can supplement epidemiological findings to provide further insight into the relationship between V . cholerae isolates and epidemic populations , identify clusters , establish phylogeny , and track bacterial transmission ., We describe our findings country by country , from Benin to Mauritania ., Databases of all suspected cholera cases were collected from the epidemiological units of Benin , Togo , Ghana , and Guinea ., National databases comprised weekly case/death numbers at the district level ( Ghana and Togo ) , commune level ( Benin ) , or prefecture level ( Guinea ) since 2009 , according to the WHO cholera case definition ( S1 Text ) ., For Ghana ( Accra ) , Togo ( Lomé ) , Benin ( Atlantique and Littoral/Cotonou ) , and Guinea ( Conakry ) , we also analyzed the cholera case line lists , which include data on age , sex , clinical outcome , and residence ., We obtained approval from the Ministry of Health ( MoH ) of each country to use these databases for epidemiological , research , and publication purposes ., The line lists were anonymized and cleaned prior to analysis ., Daily-accumulated rainfall data for the Greater Accra Region ( GAR ) were obtained from satellite estimates ( TRMM_3B42RT_DAILY . 007 ) from NASA ( http://disc . gsfc . nasa . gov/precipitation/tovas ) ., For each field visit carried out in Ivory Coast , Ghana , Togo , and Benin , we first organized and coordinated the study with UNICEF and WHO ( for Ivory Coast ) , through whom we were put in contact with national authorities to obtain official access to databases and V . cholerae strains ( when possible ) ., Initial field visits involved contact with these local UNICEF and/or WHO offices as well as national public health ( surveillance and laboratory ) authorities ., To identify cholera hotspots and sites that may play a role in cholera diffusion , we first analyzed outbreak histograms and mapped weekly cholera cases ., Field visits were then performed in the identified locales ( e . g . , sites that were repeatedly affected by cholera outbreaks or sites of an initial outbreak ) ., During field visits , and accompanied by local counterparts , we met with local surveillance departments , laboratories ( for information concerning lab-based case confirmation ) , and health facilities ., Information was collected concerning how cholera was contracted as well as possible links with other cases ., We also evaluated local WASH ( Water , Sanitation , and Hygiene ) conditions ., Field visits were performed in Ivory Coast ( 12/2013; SM and RP ) ; Ghana , Togo and Benin ( 11-12/2014; SM , PC , and RP ) ; Guinea and Sierra Leone ( 08-09/2012; SR ) ., See S1 Text for additional details concerning the field visit study protocol ., The study was approved by the MoH of each country where field visits were carried out ., The protocol was further approved by the Ghana Health Service Ethical Review Board according to standard procedures ., The remaining countries did not seek ethics approval as epidemic disease surveillance and response is covered by national public health laws as an integral part of the public health mandate of each MoH ., All clinical isolates studied were analyzed anonymously ., The country maps were generated using QGIS v2·8-Wien with shapefiles from DIVA-GIS ( http://www . diva-gis . org/gdata ) ., The shapefile of Accra was generated with QGIS in collaboration with the Ministry of Local Government , Division of Environmental Health , Accra ., To design effective public health strategies to combat cholera , it is critical to understand the mechanisms of cholera emergence and diffusion in a region-specific manner ., Genetic analysis of responsible strains can supplement epidemiological findings to provide further insight into the relationship between pathogenic strains and epidemic populations 21 ., Indeed , isolate genotyping is useful to differentiate between different isolates , identify clusters , establish phylogeny , and track bacterial transmission ., Lam et al . 22 have shown that MLVA represents a highly discriminatory technique to distinguish between closely related seventh pandemic isolates ., They have also emphasized that the method is best applied for outbreak investigations or to identify the source of an outbreak ., Rebaudet et al . have recently demonstrated that MLVA-based analysis of clinical V . cholerae isolates combined with an epidemiological assessment was instrumental in deciphering the origin of the 2012 cholera epidemic in Guinea 10 ., All V . cholerae isolates were selected in a manner as to represent both the spatial and temporal evolution of each cholera epidemic in each country sampled ., A total of 173 V . cholerae O1 clinical isolates collected throughout Ghana from 2010 to 2014 were provided by the National Public Health Reference Laboratory , Accra ., The isolates were sub-cultured and transported in glycerol tubes at ambient temperature to Marseille , France ., Aliquots of the culture were directly submitted for DNA extraction ., We also analyzed three V . cholerae isolates from Senegal in 2011 following the same procedure ., The MLVA results of the Ghanaian isolates were compared with those of previously analyzed strains from Togo ( 35 isolates ) , Guinea ( 37 isolates ) , and Sierra Leone ( 9 isolates ) as previously described 10 , 23 ., DNA was extracted using a NucliSENS easyMAG platform ( bioMérieux ) ., MLVA of the isolates using an ABI PRISM 3130 Genetic Analyzer ( Applied Biosystems ) was performed using six VNTRs as described previously 23 , and the relationship between the isolates was established using the goeBURST algorithm on PHYLOViZ v1 . 1 ( http://www . phyloviz . net/ ) ., We performed a phylogenetic assessment of the core V . cholerae genome of the strains of the third wave of the seventh pandemic 24 based on genome-wide SNPs ., Isolates from Togo-2010 ( six ) , Togo-2011 ( six ) , Togo-2012 ( five ) , Ghana-2011 ( one ) , Ghana-2014 ( five ) , and Guinea-2012 ( two ) were also included in the analysis ., DNA was sequenced using a HiSeq Illumina System ( Illumina ) and analyzed as described 24 ., From 2009 to 2015 , Benin and Togo accounted for a combined average of 694 reported cholera cases annually ( Table 1 ) ., In Benin , the lakeside commune of Sô-Ava , which is directly connected to Nigeria via Lake Nokoué and Yewa River , reported cholera outbreaks every year since 2010 and was often the first and hardest-hit commune ., In Benin , Sô-Ava reported 40% of all cases in 2013 and 30 . 4% of all cases in 2014 ( S1 Fig ) ., Cotonou , the economic center of Benin , was only affected by limited cholera outbreaks in 2010 , 2011 , and 2013 , in neighborhoods characterized by fishing activity and pronounced population movement ( S1 Fig ) ., In neighboring Togo , most outbreaks in Lomé , the capital of Togo , occurred in flood zones ( Lomé D2: Adakpamè , Bè Kpota , Anfamé , and Akodéssewa ) or areas linked with fishing activity and intense population movement ( Lomé D3: Katanga ) ( S2 Fig ) ., Meanwhile , the outbreaks in the eastern Lacs Prefecture , Togo ( in Séko ) were associated with people attending traditional animist ceremonies , including those who traveled from Benin or Nigeria , as noted by health facility staff in Séko ., Outbreaks in Togo often remain limited ( S2 Fig ) ., Ghana accounted for 52 . 4% of all reported cholera cases in coastal countries from Benin to Mauritania , from 2009 to 2015 ( 51 , 333 suspected cases in Ghana / 97 , 887 total suspected cases in the 11 countries ) 2 ., Since 2011 , cholera outbreaks have significantly intensified in Accra , the capital of Ghana ., The majority ( 73 . 6% ) of cases from 2011 to 2014 were reported in the Greater Accra Region ( GAR ) ( 35 , 985 cases GAR/ 48 , 914 cases Ghana ) ( MoH ) ., In 2012 and 2014 , the first confirmed cholera cases were detected in GAR ., By contrast , the 2011 epidemic began in late 2010 , during which the first detected cases were in Central Region ( S3 Fig ) ., The 2010/2011 epidemic then intensified in GAR in January 2011 , as displayed in Fig 1 ., Fig 1 displays the sharp increase in cases at the onset of each epidemic , indicating a rapid early expansion of the bacterium within Accra ., Strikingly , from the end of 2012 through mid-2014 , Ghana experienced an 18-month lull in cholera cases , despite typical rainfall ., All 20 suspected cholera case samples in 2013 tested negative for V . cholerae ., After this significant lull , Ghana experienced the largest epidemic ( 28 , 944 cases in 2014 ) since 1991 2 ., Notably , the strains causing this large outbreak in 2014 were closely related with strains present in Togo in 2010 and 2011 , as demonstrated by the MLVA and phylogeny data described in further detail below ., Once outbreaks erupted in Accra , cholera rapidly diffused throughout the majority of the city ., This rapid spatial diffusion pattern was observed during the onset of the 2011 , 2012 , and 2014 epidemics ., Many Accra neighborhoods were severely affected by cholera each year ., However , certain nearby residential areas remained largely cholera-free , despite outbreaks in adjacent neighborhoods ( Fig 2 ) ., Once cholera erupted in Accra , outbreaks spread to other districts in Ghana several weeks later ., According to health facility staff at the hospital in Ho ( Volta Region ) , the 2014 index case in Ho had recently traveled from Accra , where an outbreak was ongoing at the time ( S4 Fig ) ., Field visits revealed that water distribution was often interrupted for several days in many Accra neighborhoods ., The majority of water network pipes were visibly damaged and running along the ground through roadside gutters ., Residents without access to proper latrine facilities perform open defecation into these gutters ., In the Greater Accra Metro Area , access to improved sanitation facilities is limited ., Only 5 . 7% of households in Greater Accra have access to latrines that flush into a piped sewer system , while 35 . 1% of Accra households use latrines that flush into a septic tank ., Furthermore , Accra households lacking improved sanitation facilities ( 14 . 8% ) are forced to use pan latrines ( buckets that are then dumped into the roadside gutters ) or “flying toilets” ( defecation into a plastic bag which is then literally thrown away ) 25 ., We thus hypothesize that ground water and human waste could seep into broken pipes , especially during the frequent water shortages , thereby allowing V . cholerae to enter the water network and spread when water pressure is restored ., Increased rainfall markedly exacerbated this effect ( Fig 1 and Fig 2 ) ., Additional studies should further investigate the role that the Accra water network plays in cholera outbreaks to confirm this hypothesis ., From 2009 to 2010 , Ivory Coast was largely unaffected by cholera ( 37 suspected cases ) 2 ., However , an epidemic erupted in Abidjan in January 2011 following the post-election crisis of November 2010 and a collapse in the health and sanitation systems 26 ., The 2011 epidemic was responsible for 1 , 261 cases 2 ., We found that a new outbreak emerged in May 2012 in Sud Comoé , adjacent to Jomoro District in Ghana , where an outbreak was ongoing ., As observed in Ghana , Ivory Coast also experienced a complete lull in cholera in 2013 and early 2014 ., Each of the 56 suspected cases in 2013 tested negative for V . cholerae 27 ., This lull was interrupted in early October 2014 , when an outbreak occurred in Abidjan with the arrival of ill Ghanaian fishermen ( S5 Fig ) 28 ., Liberia has reported 4 , 133 suspected cholera cases from 2009 to 2015 2 , which includes only 44 cases in 2014 and zero cases in 2015 ., However , no cholera-related deaths were reported from 2010 to 2013 , and only two deaths ( 1 , 070 cases ) were reported in 2009 ., Following a three-year lull in the incidence of cholera , both Sierra Leone and Guinea experienced a trans-border epidemic in 2012 , with 22 , 932 and 7 , 351 cases , respectively 10 ., The two epidemics progressed following a very similar pattern ., In Guinea , the outbreak started in February on Kaback Island with a fisherman traveling from Sierra Leone ., In Sierra Leone , possible events of V . cholerae importation by fishermen travelling from Liberia and Ghana have been reported 29 ., During the rainy season , cholera exploded in the capitals , which recorded over half of the total cases ( Freetown , 52%; Conakry , 64% ) 10 , 29 ., As cholera rates declined in Sierra Leone and Guinea , they rose in Guinea-Bissau 29–31 , which also followed a near three-year lull ., The country reported 3 , 068 cases in 2012 and 969 cases in 2013 ., Eighteen and zero cases were reported in 2014 and 2015 , respectively 2 ., The number of cholera cases reported in The Gambia , Senegal , and Mauritania has been very low since 2009 to present ., The Gambia has not reported a single suspected cholera case since 2008 ., Likewise , Mauritania has not reported cholera cases since 2008 , with the exception of 46 cases in 2011 ., Since 2009 , Senegal has reported only 13 suspected cholera cases 2 ., We performed MLVA of 255 clinical V . cholerae isolates from Ghana , Togo , Guinea , Sierra Leone , and Senegal ., Two environmental isolates from Guinea 2012 were also included ( S1 Table ) ., Interestingly , the Minimum Spanning Tree ( MST ) shows that the 2010/2011 isolates from Ghana were related to those that seeded the epidemic in Guinea and Sierra Leone in 2012 ., The MST also demonstrates that the 2011 , 2012 , and 2014 epidemics in Ghana were due to three distinct V . cholerae MLVA-type clusters ., The three clinical isolates from Senegal in 2011 displayed an identical MLVA type , closely related to strains from Togo in 2011 and 2012 , which may represent imported cases from farther south in West Africa ( Fig 3 ) ., When the Ghana and Togo strains were included in the alignment of the seventh pandemic V . cholerae isolates 24 , the Ghana 2011 and closely linked strains from Togo ( 2010 and 2011 ) grouped with the Guinea 2012 strains on the third wave of the current pandemic ., By contrast , Ghana 2014 and other Togo strains clustered together in a separate clade , genetically distinct from the Ghana 2011 cluster ., Some of the strains from Togo in 2012 clustered together with Ghana strains from 2012 , which were more closely related to the Ghana 2014 strains on the MST ( Fig 4 ) ., From 2009 to 2015 , we found that Accra , the capital of Ghana , reported the highest number of cholera cases during the study period among all cities included in this study ., During the period 2009 to 2015 , we found that one major wave of cholera outbreaks spread from Accra in 2011 northwestward to Sierra Leone , Guinea , and likely Guinea-Bissau in 2012 ( S5 Fig ) ., Genetic analysis showed that the 2012 isolates from Guinea and Sierra Leone clustered with those collected in Ghana in 2011 ., MLVA also demonstrated that the V . cholerae strain responsible for the epidemic in Ghana , which started in 2010 and spilled over into 2011 , was already present in Togo and thus likely shared a common ancestor with strains from Togo in 2010 ., The MST and whole-genome sequence results indicate the presence of different V . cholerae populations in Togo , one group that appears to give rise to the 2011 Ghana strains , while the other group was closely related to the strain that triggered the cholera epidemic in Ghana in 2014 ., We noted that other cities ( Conakry and Freetown ) also appeared to function as amplifiers of cholera , when cases were present and rainfall increased ., Strikingly , we found that many countries deemed cholera endemic in modeling studies 32 actually suffered very few outbreaks , with multi-year lull periods during which no cholera cases were detected ., Extended lulls in cholera incidence occurred despite increased rainfall , typical high temperatures , slums , and population exposure to coastal environments ., According to the WHO , cholera endemic countries have been defined as countries in which confirmed cholera cases were reported in at least three of the five past years 32 ., In fact , some of the countries in our study do not fit the definition of endemic , as at least one case must be confirmed ., To improve the current classification methodology , we may consider repeated , lab-confirmed cholera outbreaks in a country to indicate endemicity , keeping in mind that this status is fluid and may change as sanitation and water conditions improve ., Our findings and independent reports indicate that the Accra water network may play a role in rapid diffusion of cholera throughout a majority of the city , when cholera cases are present in neighborhoods where sanitary facilities and access to safe water are lacking ., A study in Accra ( in Osu Klottey Sub Metro ) has shown that drinking community pipe-borne water ( OR = 2 . 15 ) was associated with cholera in 2012 33 ., Furthermore , a separate study has revealed unsuitable residual chlorine levels and the regular presence of fecal coliform in the Accra network water 34 ., During the period 2009 to 2015 , our findings show that one major wave of cholera epidemics spread northwestward from Accra in 2011 to Sierra Leone and Guinea in 2012 ., As Ghanaian strains from previous years fail to reappear during subsequent epidemics , we hypothesize that epidemics affecting Accra may likely originate due to imported cases from a nearby cholera hotspot ., Neighboring Nigeria represents one of the major cholera foci in the world 35 , with 130 , 007 cases reported from 2009 to 2015 2 ., The country is also a likely source of outbreaks in the Lake Chad Basin 35 ., The lull in Ghana during 2013 paralleled a relatively low number of cholera cases reported in Nigeria in 2012 and 2013 ., The 2014 Ghanaian epidemic coincided with an epidemic rebound in Nigeria 2 ., Furthermore , intense commercial activity via road and boat may represent a major pathway by which cholera is imported from outbreaks in Nigeria to susceptible waterfront communities in Benin ., We hypothesize that the detection of closely related strains in Ghana and Togo throughout the study period indicates that this strain may share a common ancestor with strains responsible for persisting outbreaks in a neighboring cholera hotspot such as Nigeria ., A phylogenetic analysis of clinical isolates has shown that the current pandemic is characterized by successive global clonal expansion of three waves of closely related serotype O1 El Tor lineages emanating from the Bay of Bengal 24 ., A recent study in Science by Weill et al . has analyzed genomic data from 1070 V . cholerae O1 isolates , across 45 African countries and over a 49-year period , to show that past epidemics were attributable to a single expanded lineage ., Weill et al . show that seventh cholera pandemic V . cholerae El Tor sub-lineages from Asia were repeatedly introduced into West Africa as well as East and Southern Africa ., Epidemic waves then propagated regionally over a period of several years , which correlates with our observations 36 ., Their findings strongly suggest that human factors play a much more important role in cholera dynamics and the long-term spread and maintenance of V . cholerae in Africa than environmental factors 36 ., Further whole-genome sequence analysis of strains circulating in West Africa would greatly enhance our understanding of V . cholerae transmission pathways in the region ., Concerning the study limitations , although the databases used in this study are reliable thanks to cholera surveillance standardization throughout the region ( S1 Text ) , the system does not guarantee detection of every single sporadic or unique imported cholera case ., However , cholera cases are quickly detected once an outbreak occurs , especially when cholera-related deaths result ., Many countries deemed cholera endemic have experienced several year-long lulls in cholera outbreaks ( when all suspected cholera cases are confirmed negative upon culture ) ., Nevertheless , countries considered cholera endemic anticipate outbreaks and perform culture-based tests on samples derived from suspected cholera cases each year ., Given the capacity of the countries concerned ( which can vary over time and geographically within a country ) , it is not possible to bacteriologically confirm every suspected case , which could have an effect on the validity of the results ., Similar issues concerning surveillance capacity may have affected the completeness of suspected cholera case data collection over time ., Nevertheless , our investigations have found that the lack of cholera during a lull period is not due to inadequate surveillance or laboratory incompetence but simply due to the absence of cholera cases in the country ., The nonexistence of cholera outbreaks during the recent Ebola crisis in Guinea , Sierra Leone , and Liberia highlights the complete lull in cholera cases despite a heightened disease surveillance system ., Furthermore , all isolates analyzed via MLVA were not submitted for whole-genome sequencing due to inadequate shipping conditions , which yielded sufficient DNA for only PCR-based MLVA ., V . cholerae isolates from countries such as Ivory Coast , Benin , Guinea-Bissau , and Liberia were not included as the concerned national laboratories did not provide isolates for this investigation ., We also acknowledge that our study would be strengthened by sequence-based phylogenic results of all samples assessed via MLVA , and thus efforts are currently underway to complete our panel of V . cholerae isolates ., Furthermore , we were not able to include all isolates for sequence analysis , as certain samples lacked sufficient concentrations of DNA for the technique , thus introducing a possible bias due to incomplete selection of samples for whole-genome sequencing ., To prevent expansion of cholera outbreaks in the analyzed region of West Africa , epidemiological surveillance should be enhanced in identified vulnerable zones , such as Accra ., In vulnerable areas , improved monitoring of the drinking water supply as well as ensuring water quality and proper chlorination would mitigate epidemics and perhaps stop cholera propagation ., Based on our findings , we hypothesize that once cases arrive in the urban settings with poor sanitation facilities ( as observed in Accra , Conakry , and Freetown ) , increased rainfall provokes the infiltration of human waste , and therefore toxigenic V . cholerae , into the water network via damaged water pipes , thus promoting a rapid increase in cholera incidence ., These findings may serve as a guide to better target cholera prevention and control interventions in the identified cholera hotspots in West Africa ., Our study also highlights the value of this type of study combining epidemiological and molecular data to gain insight into the dynamics of cholera , especially in Africa . | Introduction, Methods, Results, Discussion | The countries of West Africa are largely portrayed as cholera endemic , although the dynamics of outbreaks in this region of Africa remain largely unclear ., To understand the dynamics of cholera in a major portion of West Africa , we analyzed cholera epidemics from 2009 to 2015 from Benin to Mauritania ., We conducted a series of field visits as well as multilocus variable tandem repeat analysis and whole-genome sequencing analysis of V . cholerae isolates throughout the study region ., During this period , Ghana accounted for 52% of the reported cases in the entire study region ( coastal countries from Benin to Mauritania ) ., From 2009 to 2015 , we found that one major wave of cholera outbreaks spread from Accra in 2011 northwestward to Sierra Leone and Guinea in 2012 ., Molecular epidemiology analysis confirmed that the 2011 Ghanaian isolates were related to those that seeded the 2012 epidemics in Guinea and Sierra Leone ., Interestingly , we found that many countries deemed “cholera endemic” actually suffered very few outbreaks , with multi-year lulls ., This study provides the first cohesive vision of the dynamics of cholera epidemics in a major portion of West Africa ., This epidemiological overview shows that from 2009 to 2015 , at least 54% of reported cases concerned populations living in the three urban areas of Accra , Freetown , and Conakry ., These findings may serve as a guide to better target cholera prevention and control efforts in the identified cholera hotspots in West Africa . | We analyzed cholera epidemics from Benin to Mauritania , during 2009 to 2015 , and performed a series of field visits as well as molecular epidemiology analyses of V . cholerae isolates from most recent epidemics throughout West Africa ., We found that at least 54% of cases concerned populations living in the three urban areas of Accra , Freetown , and Conakry ., Accra , Ghana represented the main cholera hotspot in the entire study region ., Our findings indicate that the water network system in Accra may play a role in the rapid diffusion of cholera throughout the city ., As observed in Accra , Conakry , and Freetown , once cholera cases arrive in overpopulated urban settings with poor sanitation , increased rainfall facilitated the contamination of unprotected water sources with human waste from cholera patients , thus promoting a rapid increase in cholera incidence ., To more efficiently and effectively combat cholera in West Africa , these findings may serve as a guide to better target cholera prevention and control interventions . | guinea, medicine and health sciences, pathology and laboratory medicine, pathogens, vibrio, tropical diseases, geographical locations, microbiology, bacterial diseases, vibrio cholerae, benin, neglected tropical diseases, bacteria, bacterial pathogens, africa, togo, infectious diseases, cholera, medical microbiology, microbial pathogens, sierra leone, people and places, ghana, biology and life sciences, organisms | null |
journal.pcbi.1000001 | 2,008 | Genome Landscapes and Bacteriophage Codon Usage | The genomes of most organisms exhibit significant codon bias—that is , the unequal usage of synonymous codons ., There are longstanding and contradictory theories to account for such biases ., Variation in codon usage between taxa , particularly within mammals , is sometimes attributed to neutral processes—such as mutational biases during DNA replication , repair , and gene conversion 1–4 ., There are also theories for codon bias driven by selection ., Some researchers have discussed codon bias as the result of selection for regulatory function mediated by ribosome pausing 5 , or selection against pre-termination codons 6 , 7 ., However , the dominant selective theory of codon bias in organisms ranging from E . coli to Drosophila posits that preferred codons correlate with the relative abundances of isoaccepting tRNAs , thereby increasing translational efficiency 8–13 and accuracy 14 ., This theory helps to explain why codon bias is often more extreme in highly expressed genes 15 , or at highly conserved sites within a gene 14 ., Translational selection may also explain variation in codon usage between genes selectively expressed in different tissues 16 , 17 ., However , recent work suggests that synonymous variation , particularly with respect to GC content , affects transcriptional processes as well 18 ., The codon usage of viruses has also received considerable attention 19 , 20 , particularly in the case of bacteriophages 21–26 ., Most work along these lines has focused on individual phages , or on the patterns of genomic codon usage across a handful of phages of the same host ., Here , we provide a systematic analysis of intragenomic variation in bacteriophage codon usage , using 74 fully sequenced viruses that infect a diverse range of bacterial hosts ., Motivated by energy landscapes associated with DNA unzipping 27 , 28 , we develop a novel methodological tool , called a genome landscape , for studying the long-range properties of codon usage across a phage genome ., We introduce a series of randomization tests that isolate different features of codon usage from each other , and from the amino acid sequence of encoded proteins ., Thirty-three of the phages in our analysis are shown to exhibit non-random variation in synonymous GC content , as well as non-random variation in codons adapted for host translation , or both ., Additionally , we demonstrate that phage genes encoding structural proteins are significantly more adapted to host-preferred codons compared to non-structural genes ., We discuss our results in the context of translational selection and lateral gene transfer amongst phages ., We start by introducing the concept of a genome landscape , which provides a simple means for visualizing long-range correlations of sequence properties across a genome 29 ., A genome landscape is simply a cumulative sum of a specified quantitative property of codons ., The calculation of the cumulative sum is straightforward , and it consists of scanning over the genome sequence one codon at a time , gathering the property of each codon , and summing it with the properties of previous codons in the genome sequence ., Similar cumulative sums are used in solid-state physics for , e . g . , the calculation of energy levels 30 ., In the case of the GC3 landscape , we have ( 1 ) where ηGC3 ( m ) equals one or zero , depending upon whether the mth codon ends in a G/C or A/T , respectively ., Note that we subtract the genome-wide average GC3 content , , so that FGC3 ( 0 ) =\u200aFGC3 ( N ) =\u200a0 , where N is the length of the genome ., In other words , we convert the genome codon sequence into a binary string of 1s and 0s according to whether each codon is of type GC3 or AT3 , and we cumulatively sum this sequence to compute FGC3 ( m ) ., The interpretation of a GC3 landscape is straightforward ., Regions of the genome whose landscape exhibits an uphill slope contain higher than average GC3 content , whereas regions of downhill slope contain lower than average GC3 content ., The genome landscape provides an efficient visualization of long-range correlations in sequence properties across a genome , similar to the techniques introduced by Karlin 31 ., Traditional visualizations of GC3 content involve moving window averages of %GC3 over the genome 32 ., In order to compare these techniques with the landscape approach , we focus on the E . coli phage lambda as an illustrative example ., Figure 1A shows the lambda phage GC3 landscape above its associated “GC3 histogram” ., The histogram shows the GC3 content of each gene , and the width of each histogram bar reflects the length of the corresponding gene ., Thus , the gene-by-gene histograms mimic a sliding window average view of nucleotide content across the genome , but focus on the contributions of individual genes to these sequence properties ., Figure 1A reveals a striking pattern of lambda phage codon usage: the genome is apparently divided into two halves that contain significantly different GC3 contents 33 , 34 ., The large region of uphill slope on the left half of the GC3 landscape reflects the fact that the majority of the genes in this region contain an excess of codons that end in G or C . This trend is also reflected in the GC3 histogram bars , which are higher than average in the left half of the genome ( Figure 1 ) ., It is clear that genome landscapes contain the same information as gene-by-gene histograms ., However , as has been noted before 29 , genome landscapes also represent a powerful visualization tool that emphasizes genome-wide trends in sequence properties ., As we demonstrate below , gene-by-gene histograms offer a mechanism by which to quantify these trends , while the landscapes offer striking views of these trends that can aid in their interpretation ., In addition , GC-landscapes are directly useful for modeling physical properties of DNA unzipping 28 ., Genome landscapes also provide a natural means of evaluating whether or not features of codon usage are due to random chance ., Under a null model in which the η ( i ) s above are chosen as independent random variables with var ( η ( i ) ) =\u200a〈η ( i ) 2〉−〈η ( i ) 2〉\u200a=\u200aΔ , one can show ( see Methods ) that the standard deviation of F ( GC3 , m ) is ( 2 ) This quantity is shown as a purple band in Figure 1 ., For η ( i ) s chosen to be 0 or 1 at random , ΔGC3\u200a=\u200a1/4 and the maximum width is obtained at m\u200a=\u200aN/2 ., Since the scale of variation across the lambda phage GC3 landscape is much greater than its expectation under the null , we can conclude that the distribution of G/C versus A/T ending codons is highly non-random in the lambda phage genome ., We can also gain intuition about the degree of non-randomness in the GC3 landscape by considering what would happen if the lambda phage genome were to accumulate random synonymous mutations ., Figure 2A shows snapshots of the lambda GC3 landscape as we simulate synonymous mutations to the genome ., Between each snapshot , N synonymous mutations were introduced by picking a codon at random along the genome , and then choosing a new synonymous codon at random according to the global lambda phage codon distribution ., By preserving the global codon distribution in each synonymous variation of the genome , this procedure inherently controls for any mutational bias or other source of global codon usage bias that may be present in the phage genome nucleotide content ., The same is true for all randomization tests discussed in this paper ., As more mutations are introduced , the GC3 landscape of the synonymously mutated lambda genome approaches the purple band , indicating that the GC3 pattern in the real lambda phage genome is highly non-random ., The procedure of producing a genome landscape can be applied to other properties of codon usage ., In addition to GC3 , we will study patterns in the Codon Adaptation Index ( CAI ) ., CAI measures the similarity of a genes codon usage to the ‘preferred’ codons of an organism 35—in this case , the host bacterium of the phage under study ., Every bacterium has a preferred set of codons defined as the codons , one for each amino acid , that occur most frequently in genes that are translated at high abundance ., These genes are often taken to be the ribosomal proteins and translational elongation factors 35 ( see Methods ) ., In order to calculate CAI , the preferred codons are each assigned a weight w\u200a=\u200a1 ., The remaining codons are assigned weights according to their frequency in the highly-translated genes , relative to the frequency of the w\u200a=\u200a1 codon ., The CAI of a gene is defined as the geometric mean of the w-values for its codons ( 3 ) where wi is the w-value of the ith codon , and M is the length of the gene ., This quantity can be re-written as ( 4 ) The latter formulation is more useful for calculating genome landscapes , because the argument of the exponential function is now a sum of the logs of the w-values ., Therefore , we define the CAI landscape as ( 5 ) where ηCAI ( m ) =\u200aln ( wm ) ., The CAI landscape for lambda phage is shown in Figure 1B , along with the CAI histogram of lambda phage ., For the CAI histograms , the height of each bar represents the CAI value of that gene ( Equation 3 ) ., As in the case with the GC3 landscape , we find that the lambda phage CAI landscape corresponds closely to the CAI histogram , but it offers a more striking global view of the long-range CAI structure in the lambda phage genome ., One contiguous half of the lambda phage genome exhibits elevated CAI , whereas the other half exhibits depressed CAI ., The observed CAI landscape lies far outside the purple band in Figure 1 , calculated according to Equation 2 , indicating that the pattern of CAI across the lambda phage genome is non-random ., However , the purple band is wider for the CAI landscape than for the GC3 landscape , because the variance in the ln ( wi ) s , ΔCAI , is greater than ΔGC3 ., The GC3 and CAI landscapes for lambda phage are highly correlated with each other ( Figure 1 ) ., In particular they both have large uphill regions on the left-hand side of the genome , indicating a region containing codons with elevated GC3-content and CAI values , compared to the genome average ., It is possible that the observed correlation between the GC3 and CAI landscapes could be caused by the conflation between high CAI and GC3 in the preferred E . coli codons , as we discuss below ., We note that the genes in the region of elevated CAI primarily encode the highly translated structural proteins that form the capsid and tail of the lambda phage virions ., This pattern suggests the hypothesis that , because of the need to produce structural genes in high copy number during the viral life cycle , structural genes preferentially use codons that match the hosts preferred set of codons ., We will explore this translational-selection hypothesis in greater detail below ., The previous section illustrated that the codon usage across the lambda phage genome is highly non-random with respect to both GC3 and CAI ., In this section we quantify this statement , and we focus on aspects of lambdas codon usage patterns that are independent of the amino acid sequences of the encoded proteins ., Since we are interested in studying the patterns of synonymous codon usage , it is important that we control for the amino acid sequence of encoded proteins ., Phages utilize a diverse spectrum of proteins , ranging from those that form the protective capsid for nascent progeny , to those encoding for the tail and tail fibers , to those that regulate the switch between lytic or lysogenic infection pathways ., As with other organisms , phage proteins have been selected at the amino acid level for function and folding ., Some portion of a phages codon usage is surely influenced by selection for amino acid content ., We can construct a simple randomization test to interrogate the potential influence of the amino acid sequence on the GC3 and CAI landscapes of lambda phage ., In this test , we generate random genomes that have the exact same amino acid sequence as lambda phage , but shuffled codons , such that the genome-wide , or global , codon distribution is preserved in each random genome ( see Methods ) ., As summarized in Table 1 , we refer to this test as the ‘aqua’ randomization test ., For each of the randomized genomes , we calculate GC3 and CAI landscape ., Similar to a recent randomization method 36 , we then compare the observed landscape of the actual genome to the distribution of landscapes generated from the randomized genomes ., Figure 3 shows the results of this comparison , with the observed landscapes plotted as black lines , and the mean±one and two standard deviations of random trials shown in dark and light aqua , respectively ., As the figures show , the observed landscapes lie in the far extremes of the randomized distributions – indicating that the amino acid sequence of the lambda phage genome does not determine the extraordinary features of the observed landscapes ., It is also instructive to query the influence of amino acid content on codon usage in each gene individually ., The histogram view of these randomization tests allows us to ask this question precisely ., Because the amino acid sequence is preserved exactly across the genome , each histogram bar in Figure 3 can be considered as its own randomization test , one for each gene ., The position of the horizontal black bar reflects the actual codon usage of each gene , and it can be compared to the distribution of random trials in order to compute a quantile for each gene: ( 6 ) Note that we have defined two quantiles , q> and q< , that describe the proportion of random trials strictly less or strictly greater than the observed data ., These two quantities sum to a values less than one ( and equal to one if there are no ties ) ., A value of q>>0 . 5 signifies that the observed statistic ( e . g . GC3 or CAI ) is greater than most of the random trials ., Associated with each of these quantiles is a p-value quantifying whether the observed gene sequence has significantly different codon usage than the random trials: p<\u200a=\u200a1−q< and p>\u200a=\u200a1−q> ., If either one of these p-values is low , it signifies that the GC3 ( or CAI ) content of the gene is significantly different than the genomic average , controlling for the amino acid sequence of the gene ., p< tests for significantly depressed GC3 ( or CAI ) in a gene; and p> tests for significantly elevated GC3 ( or CAI ) in a gene ., We will use these p-values , which arise from the ‘aqua’ randomization test , in two ways ., Since we are interested in studying the effects of synonymous codon usage alone , we first wish to filter out any genes whose codon usage does not significantly deviate from random , given the amino acid sequence ., Therefore , in the subsequent gene-by-gene analyses reported in this paper , we retain only those genes whose quantiles fall in the extreme 5% of random trials ., That is , we only keep those genes for which or ., These genes are said to ‘pass’ the aqua test , and they are unshaded in Figure, 3 . We also use the gene-by-gene p-values to quantify the degree to which codon usage is independent of amino acid sequence across the genome as a whole ., To do so , we combine all the gene-by-gene p-values into an aggregate p-value for the entire genome , paqua , using the method of Fisher 37 ., We calculate the combined p-value by summing the logs of twice the minimum of each gene-specific p-value ( 7 ) where represents the aqua p<-value for gene i , and k is the number of genes in the genome ., It is well known that faqua is chi-squared distributed with 2k degrees of freedom 37 ., Thus , the combined p-value for the entire genome , , where is the cumulative chi-squared distribution with 2k degrees of freedom ., In the case of lambda phage , we find for GC3 and for CAI ., Thus , we conclude that the neither the GC3 nor the CAI patterns across the lambda phage genome are determined by the genomes amino acid sequence ., In the following sections we will use the aqua test ( see Table 1 ) and its associated gene-by-gene and combined p-values as a control to verify that features of codon usage are not driven by the amino acid sequence ., Depending upon the preferred codons of the host species , the effect of selection for high CAI in a viral gene is not necessarily independent from the effect of selection for other features of viral codon usage , such as high GC3 ., For example , codons with high CAI values associated with a given host may be biased towards high GC3 values as well ( see Figure 4 ) ., It is important , therefore , to disentangle the effects of selection for CAI versus selection for GC3 , in order to determine which one of these forces is responsible for the non-random patterns of codon usage observed in the lambda genome ., The weights used to compute CAI for E . coli are shown in Figure, 4 . The 61 codons are placed into one of four groups according to whether they are GC3 or not ( red or blue , respectively ) , and whether they have high CAI or not ( dark or light , respectively ) ., High CAI is determined by an arbitrary cutoff of w≥0 . 9 ., As this table demonstrates , the set of preferred codons in E . coli is slightly biased towards GC-ending codons ( 58% ) ., The GC bias of preferred codons , although slight , could conflate the results of selection for CAI versus GC3 in phages that infect E . coli , such as lambda ., We therefore introduce another randomization test that allows us to disentangle patterns of CAI content from patterns of GC3 content ., Similar to the aqua randomization test described above , we draw random phage genomes such that the amino acid sequence is conserved , but we add the additional constraint of conserving the exact GC3 sequence as well ( see Methods ) ., For example , at a site containing a GC3 codon for leucine , in our random trials we only allow those leucine codons terminating in G or C . By comparing the observed landscapes of the genome with the distribution of randomly drawn landscapes , we can isolate the features of codon usage driven by CAI , independent of GC3 and amino acid content ., We refer to this randomization procedure at the ‘orange’ randomization test ( Table 1 ) ., Conversely , we also wish to assess the strength of patterns in GC3 content , independent of CAI and amino acid content ., The appropriate randomization procedure in this case requires that we constrain the amino acid sequence and the sequence of codon CAI values while allowing GC3 to vary ., However , because CAI values are not binary , CAI cannot be constrained exactly while still allowing for enough variability to produce a meaningful randomization test ., Thus , we introduce a binary version of the CAI measure , called BCAI , that is qualitatively the same as and , for our purposes , interchangeable with CAI ., The BCAI w-value for a codon is defined to be 0 . 7 if the codon is high CAI , and 0 . 3 if the codon has low CAI ., High CAI is defined by the threshold of w≥0 . 9 ( see Figure 4 ) ., The threshold value w≥0 . 9 is arbitrary , and our results are robust to changing this threshold ( see Figures S1 and S2 ) ., Our use of the term ‘binary’ here refers to the binary classification scheme and not the particular values of BCAI ., The actual values assigned for BCAI are arbitrary , for the most part , and have no effect on our results ., Nevertheless , we cannot assign low BCAI a value of zero , because this value would be problematic when included in the geometric averaging procedure , or when computing the logarithm of w-values for BCAI landscapes ., BCAI provides a useful surrogate for CAI because its values are binary , thereby allowing us to constrain a genes amino acid sequence and BCAI sequence exactly , while varying GC3 content in random trials ., The BCAI landscapes and histograms are calculated in the same way as CAI landscapes and histograms , except using BCAI w-values ., As expected , the BCAI landscape of a genome is qualitatively similar to its CAI landscape ( compare Figures 5B and 3B ) , and the two landscapes are highly correlated ( e . g . r\u200a=\u200a0 . 72 for lambda phage ) ., Thus BCAI is interchangeable with CAI for the purposes of our randomization tests ., Figure 5 shows the results of the two randomization tests outlined above: the ‘green’ test that compares the observed GC3 landscape to a distribution of random trials constraining the amino acid sequence and the BCAI sequence; and the ‘orange’ test that compares the observed BCAI landscape to a distribution of random trials constraining the amino acid sequence and the GC3 sequence ., Our convention for naming these two tests is summarized in Table 1 ., As seen in Figure 5A , the observed GC3 landscape lies significantly outside of the random trials that preserve amino acid sequence and BCAI sequence ., Combining the gene-by-gene p-values for this test , we find – indicating that the lambda phage genome as a whole has non-random GC3 variation independent of amino acid and CAI ( actually , BCAI ) sequence ., Conversely , Figure 5B shows that the BCAI landscape contains non-random features when controlling for both GC3 and amino acid sequence ( ) ., In other words , the lambda phage genome exhibits highly non-random patterns of both GC3 and CAI codon variation , independent of one another and independent of the amino acid sequence ., In the sections above we have demonstrated and quantified highly non-random patterns of GC3 and CAI codon usage variation across the lambda phage genome ., We have also demonstrated that these trends are independent of one another ., In this section , we will extend our analysis to a large range of diverse phages ., In this section we consider all sequenced phages that infect E . coli , Pseudomonas aeruginosa or Lactococcus lactis as their primary host ., The latter two hosts were chosen because of they contain unusually extreme GC3 content: 88 %GC3 for P . aeurginosa and 25 %GC3 for L . lactis , genome-wide ., The extreme GC3 content of these hosts give rise to opposing relationships between high CAI and GC3 – as indicated schematically in Figure 6 ., In particular , P . aeruginosa strongly favors GC3 in high-CAI codons ( 94% ) , and L . lactis strongly favors AT3 in high-CAI codons ( 72% ) ., Thus , these three hosts span a large spectrum of relationships between CAI and GC3 ., Since our randomization tests constrain amino acid and BCAI exactly ( the ‘green’ test ) , and amino acids and GC3 exactly ( the ‘orange’ test ) , we can control for any possible conflation between GC3 and CAI trends ., Thus , the randomization tests are equally applicable to all of the phage genomes , regardless of their host ., We performed the aqua , green , and orange randomization tests on the 45 phages of E . coli , 12 phages of P . aeruginosa , and 17 phages of L . lactis whose genomes have been sequenced ( see Methods ) ., In the first step of our analysis , we removed any phages which failed either the aqua GC3 or aqua CAI tests , because the codon usage of such genomes are influenced by their amino acid sequence ., A phage was said to pass these two control tests if its Fisher combined p-values for both aqua GC3 and aqua CAI were significant ., The significance criterion for each test is pcombined<5%/74 , which incorporates a Bonferroni correction for multiple tests ., With this cutoff , 50 of the initial 74 phages passed the aqua control tests ., Figure 7 shows results of these tests for several example genomes ., P2 , a temperate phage , and T3 , a non-temperate phage both infect E . coli and both pass the control tests and exhibit significant ‘orange’ and ‘green’ results , as does D3112 , a temperate phage that infects P . aeruginosa ., However , not all phages that pass the control test exhibit significant ‘orange’ and ‘green’ results – as evidenced by bIL286 , a temperate phage infecting L . lactis ., Figure 8 plots the distribution of combined Fisher p-values of the orange and green tests , for the 50 phages that pass the control tests ., The majority of these p-values are highly significant ., Using a Bonferoni-corrected threshold of 5%/50 , a total of 22 genomes show significance in the orange test , 29 in the green test , and 17 in both orange and green ., These results indicate that non-random patterns in codon usage are not unique to lambda phage ., Indeed , over a range of bacterial hosts and a range of phage viruses , there is apparent pressure for non-random patterns of both GC3 content and CAI content , independent of one another and independent of the amino acid sequence ., In this section , we investigate a natural hypothesis concerning the patterns of non-random CAI usage we have observed in phage genomes – namely , that these patterns may be driven by selection for translational accuracy and efficiency , which is stronger in more highly expressed proteins 9 , 21 ., Among all phage proteins , the structural proteins are the most highly expressed 38 ., The structural proteins form the protective capsid that encloses the viral genome , as well as the tail , which is often used for transmission of the phage genome to the inside of the host 39 ., These proteins must be produced in high copy number – many tens of copies of each type of structural protein needed to form each of hundreds of viral progeny 38 ., For each gene in a phage genome , we assigned a structural annotation of 1 if the gene was known to encode a structural protein and 0 otherwise ( see Methods ) ., According to the standard hypothesis of translational selection , the structural genes of phages should exhibit elevated CAI levels compared to other phage genes , since they are translated ( by the host ) in high copy numbers ., To test this hypothesis , we performed regressions between the structural annotation of phage genes and their aqua CAI and orange BCAI p-values ., In other words , we compared the structural properties of genes against their CAI content , controlling for amino acid sequence , and against their BCAI content , controlling for both amino acid sequence and GC3 sequence ., In the case of lambda phage , Figure 9 shows the results of the aqua CAI and orange BCAI randomization tests , with the structural genes highlighted ., The plot reveals a striking pattern: the vast majority of the structural proteins lie on the left half of the genome , exactly in the region where genes have elevated CAI values ., In order to quantify this association we performed ANOVAs ., Before regressing structural annotations against codon usage , we first removed the non-informative genes – i . e . genes whose codon usage are influenced by their amino acid content , as indicated by a failure to pass the aqua CAI test ., Table 3 shows the results of the regression between aqua CAI and orange BCAI p>-values versus structural annotations in lambda phage ., The results are highly significant: structural annotations explain half of the variation in CAI , even when controlling for genes amino acid sequences ( aqua , r2\u200a=\u200a56% ) as well as GC3 sequences ( orange test , r2\u200a=\u200a46% ) ., The median p>-value among structural genes is close to zero , whereas the median p>-value among non-structural genes is close to one – indicating that structural genes exhibit significantly elevated CAI values ., These highly significant results are consistent with the hypothesis of translational selection on structural proteins ., In order to examine the relationship between structural annotation and CAI across all 74 phages in our study , we performed the same ANOVA on the 1 , 309 informative genes ( i . e . genes that pass the aqua CAI randomization test ) ., Once again , Table 3 shows a highly significant relationship between structural annotation and CAI values , controlling for amino acid content and GC3 ., Thus , the tendency toward elevated CAI values in structural genes holds across all the phages in this study , despite the fact that they infect a diverse range of hosts with a wide variety of GC contents ., Similar to reports for other organisms 40 , we find a relationship between gene length and codon adaptation ., In our case , however , longer viral genes are associated with more significant p>-values in the aqua and orange tests ., However , the strength of this relationship is weak , and controlling for gene length does not affect our results on elevated CAI in structural proteins ( ANOVA p-values analogous to Table 3 are less than 10−9 after controlling for gene length ) ., In this paper , we have developed genome landscapes as a tool for visualizing and analyzing long-range patterns of codon usage across a genome ., In combination with a series of randomization tests , we have applied this tool to study synonymous codon usage in 74 fully sequenced phages that infect a diverse range of bacterial hosts ., Genome landscapes provide a convenient means to identify long-range trends that are not apparent through conventional , gene-by-gene or moving-window analyses ., Using a statistical test that compares codon usage to random trials , controlling for the amino acid sequence , we found that we found that many of the phages studied exhibit non-random variation in codon usage ., However , not all of the phages exhibit non-random variation as exemplified by phage bIL286 ( Figure 7D ) ., In light of long-standing 9 and recent 18 literature from other organisms , we have focused on two aspects of phage codon usage: variation in third-position GC/AT content ( GC3 ) and variation in the degree of adaptation to the ‘preferred’ codons of the host ( CAI ) ., Almost three-quarters of the phages in our study exhibit non-random intragenomic patterns of codon usage , even when controlling for the amino acid sequence encoded by the genome ., Almost half of such genomes also show non-random patterns of CAI when additionally controlling for the GC3 sequence ., In other words , there is substantial variation in CAI above and beyond what would be expected by random chance , given the amino acid and GC3 sequences of these genomes ., We have also compared the CAI values of phage genes to their annotations as structural or non-structural proteins ., We have conclusively demonstrated that phage genes encoding structural proteins exhibit significantly elevated CAI values compared to the non-structural proteins from the same genome ., These results hold even when controlling for the amino acid sequence and GC3 sequence of genes ., Our conclusions across a diverse range of phages are consistent with early observations on lambdas codon usage 34 , early results for T7 21 , and with the general hypothesis of translational selection , which predicts elevated CAI in genes expressed at high levels 9 , 15 , 35 ., The pattern of elevated CAI in structural proteins is particularly striking the case of lambda phage ., It is also worth noting that we find no significant relationship between a phages life-history ( i . e . temperate versus non-temperate ) and the degree to which its structural proteins exhibit elevated CAI ( see Table 6 ) ., This observation likely reflects the fact that at some point every phage , regardless of its life history , must generate certain structural proteins in high abundance – and so it is beneficial to encode such protein using the hosts translationally preferred codons ., Some of the phages examined are known to encode their own tRNA genes ., Table 5 lists the number of tRNA genes for the ten phages in this study that encode tRNA genes ., We have inspected these examples for signs that structural genes might be preferentially encoded by endogenous tRNAs , or the converse , but have concluded that the data are equivocal ., There are too few informative examples to make a strong conclusion in either direction ., Our results on translational selection in phages shed light on the nature of selection on viruses ., The standard interpretatio | Introduction, Results, Discussion, Materials and Methods | Across all kingdoms of biological life , protein-coding genes exhibit unequal usage of synonymous codons ., Although alternative theories abound , translational selection has been accepted as an important mechanism that shapes the patterns of codon usage in prokaryotes and simple eukaryotes ., Here we analyze patterns of codon usage across 74 diverse bacteriophages that infect E . coli , P . aeruginosa , and L . lactis as their primary host ., We use the concept of a “genome landscape , ” which helps reveal non-trivial , long-range patterns in codon usage across a genome ., We develop a series of randomization tests that allow us to interrogate the significance of one aspect of codon usage , such as GC content , while controlling for another aspect , such as adaptation to host-preferred codons ., We find that 33 phage genomes exhibit highly non-random patterns in their GC3-content , use of host-preferred codons , or both ., We show that the head and tail proteins of these phages exhibit significant bias towards host-preferred codons , relative to the non-structural phage proteins ., Our results support the hypothesis of translational selection on viral genes for host-preferred codons , over a broad range of bacteriophages . | Any protein can be encoded by multiple , synonymous spellings ., But organisms typically prefer one spelling over another—a phenomenon known as codon bias ., Codon bias is generally understood to result from selection for synonymous spellings that increase the rate and accuracy of protein translation ., In this work , we have examined the complete genomes of all sequenced viruses that infect the bacteria E . coli , P . aeruginosa , and L . lactis , and have found that many of these viral genomes also exhibit codon bias ., Moreover , the degree of codon bias varies across the viral genome , as visualized using a technique called a “genome landscape . ”, By comparing the observed genomes to randomly drawn genomes , we demonstrate that the regions of high codon bias in these viral genomes often coincide with regions encoding structural proteins ., Thus , the proteins that a virus needs to produce in high copy number utilize the same encoding as its host organism does for highly expressed proteins ., Our results extend the translational theory of codon bias to the viral kingdom: parts of the viral genome are selected to obey the preferences of its host . | computational biology/population genetics, computational biology/sequence motif analysis, computational biology/genomics | null |
journal.pgen.1005778 | 2,016 | Integrated Multiregional Analysis Proposing a New Model of Colorectal Cancer Evolution | Cancer is a heterogeneous disease ., Recent cancer genomics studies have revealed extensive genetic diversity among patients ., Moreover , even a clonal tumor in one patient often harbors multiple subclones ., This phenomenon is called intratumor heterogeneity ( ITH ) and is presumably generated by branching clonal evolution of cancer cells ., Understanding of ITH is clinically important , since the existence of multiple subclones presumably boosts the evolutionary adaptation of tumors against therapies , constituting a source of resistant clones 1 ., Recently , a multiregional sequencing approach , which sequences DNA sampled from geographically separated regions of a single tumor , has revealed branched evolution and ITH ., Yachida et al . 2 investigated the genomic evolution of pancreatic cancer , establishing two categories of mutations: “founder” and “progressor” mutations are present in all regions and a subset of regions , respectively ., Founder mutations are assumed to appear in the early phase of clonal evolution ., We refer to the clone that has accumulated all the founder mutations as the parental clone ( or the most recent common ancestor ) ., The parental clone then branches into subclones by accumulating progressor mutations , which shape ITH ., Several studies employing multiregional exome sequencing have revealed the occurrence of branched evolution and ITH in several other types of cancers , including clear cell renal cell carcinomas and non-small cell lung cancers ., ITH of clear cell renal cell carcinomas is characterized by parallel evolution , in which the same driver gene is independently mutated in different branches of evolutionary trees 3 ., In contrast , no evidence of parallel evolution has been reported for non-small cell lung cancer 4 , 5 ., In addition to genetic aberrations , epigenetic aberrations are also a hallmark of cancer; as for DNA methylation , a few groups have also performed multiregional epigenomic analyses 6 , 7 ., However , the types of cancers that have been subjected to multiregional analyses remain limited , and ITH of genomes and epigenomes has been poorly studied in an integrated way ., In this study , we present genetic and epigenetic analysis of ITH in a series of nine colorectal cancers ., Following multiregional sampling , we performed exome sequencing and copy number ( CN ) , methylation , and mRNA expression array profiling ., Our integrated analysis revealed not only extensive ITH , but also the evolutionary histories of the nine tumors ., Finally , we also performed computational simulation of cancer evolution , which suggested a possible evolutionary principle underlying the extensive ITH ., To study ITH in colorectal cancer , we performed genomic analysis of samples from geographically separated regions from nine colorectal tumors ( S1 Table ) ., In this study , we referred to the nine patients by the term “case” and to multiregional samples in each case by the term “sample” ., From each of the nine tumor , we obtained 5–21 multiregional samples , which were 75 samples in total , together with 9 paired normal mucosa samples ( S2 Table ) ., For two cases , samples from liver metastases were obtained ., Our multiregional exome sequencing of the nine cases found 16857 mutations in total , for an average of 58–1195 mutations per sample ( S3 Table ) ., From these values , the mutation rates for each case were estimated to be 1 . 57–20 . 2 mutations per megabase ., All cases , except for case 9 , fall in a range typical for non-hypermutated colorectal cancer 8 ., Mutational profiles obtained from the multiregional sequencing demonstrated high genetic ITH for all nine colorectal tumors ( Fig 1 ) ., Each of the multiregional mutation profiles harbored founder and progressor mutations; founder mutations are shared by all regions while progressor mutations are not ., We further divided progressor mutations into two subcategories: “unique” and “shared” mutations , which are unique to a single specific sample and shared by multiple but not all samples , respectively ., Targeted deep sequencing validated 100% ( 5068/5068 ) , 93 . 9% ( 1745/1857 ) and 95 . 4% ( 1362/1427 ) of founder , shared , and unique mutations , respectively ., We can assume that founder , shared , and unique mutations are acquired in this order during cancer evolution ., Applying the maximum parsimony method 9 to the multiregional mutation profiles allowed us to depict the evolutionary trees of the nine tumors ( Fig 2 ) ., Comparison between the evolutionary trees and geographical positions of each of the samples showed that subclones were generally separated in geographically correlated ways , demonstrating that geographical relations are maintained as the evolution of colorectal cancer proceeds ., On the other hand , our analysis of the deep sequencing data revealed that some regions in two cases harbor intermixed subclones from separated regions , which confirmed a recent finding by Sottoriva et al . ( S2 Fig ) 10 ., We found that mutations in well-known driver genes such as APC , KRAS , and FBWX7 were acquired as founder mutations during the establishment of the parental clones ( Fig 3A ) ., Pathway-level analysis also showed that founder mutations disrupted the WNT and RTK/RAS pathways , consistently with their principal roles in colorectal tumorigenesis ( S3 Fig ) ., Once the parental clones were established , these clones branched into subclones by accumulating progressor mutations ., We found that mutations in PIK3CA recurrently occurred as progressor mutations , suggesting that PIK3CA mutations are a late event in the evolution of colorectal cancer ., On the other hand , we did not find any evidence of parallel evolution , as has been observed in studies of clear cell renal cell carcinomas 3 ., We counted each category of mutation and then identified a correlation between the number of founder mutations and the age of the patients ( Fig 3B , S4 and S5 Figs ) ., Our findings are consistent with the model of founder mutations accumulating from aging , and similar correlations have also been observed in other types of solid tumors , such as pancreatic cancer and clear cell renal cell carcinomas ( S5 Fig ) ., To investigate the temporal signatures embedded in the mutations , we then compared mutational signatures between founder mutations and progressor mutations ( Fig 3C ) ., Our analysis showed that C > T transitions at CpG sites are more prominent in founder mutations than in progressor mutations ., Next , we calculated the fraction of cancer cells harboring each category of mutation from variant allele frequency ( VAF ) , read depth and CN data ( Fig 3D ) ., We found that the cancer cell fractions decreased while proceeding from founder to shared and unique mutations; that is , founder and progressor mutations tend to exist as clonal and subclonal mutations in each sample , respectively ., In our multiregional sampling , we estimated that the cell population sizes of each sample are about 106 from the amount of DNA , while those of each case as a whole are from 109 to 1010 based on tumor size ., This observation suggests that ITH is extremely extensive and the resolution of our multiregional sampling remains insufficient to reveal its totality ., This result also prompted us to estimate founder and progressor mutations in single sample sequencing data ( S6 Fig ) ; from The Cancer Genome Atlas ( TCGA ) colon and rectum adenocarcinoma exome sequencing data 8 , we obtained clonal and subclonal mutations as surrogates of founder and progressor mutations , respectively ., Using this data , we confirmed that clonal mutations are increased with patients’ ages and that they have a higher proportion of C>T transitions at CpG sites than do subclonal mutations ( Fig 3C and 3E and S7 Fig ) ., A recent pan-cancer analysis reported that C>T transitions at CpG sites are positively correlated with patients’ ages 11 ., Consistently , we confirmed that C>T transitions at CpG sites are increased with patients’ ages in the TCGA data ( S7 Fig ) ., Taken together , the TCGA data analysis supported the hypothesis that founder mutations enriched with C>T transitions at CpG sites were accumulated during aging ., In addition to the exome sequencing , we also performed single-nucleotide polymorphism ( SNP ) array-based CN profiling for all cases except case 1 ( Fig 1 and S9 Fig ) ., The multiregional CN profiles showed that amplifications of 7p , 13q , 10q , 20p , and 20q frequently occurred across all samples in multiple tumors , namely as founder CN alterations ( Fig 3A ) ., We also found that , compared with the degrees of mutational ITH , those of CN ITH were variable among cases ., For example , cases 2 , 3 , and 7 showed relatively high CN ITH , which was acquired in a manner that correlated with mutational ITH , as shown by cluster analysis ( S9 Fig ) ., In addition to founder CN alterations , we identified CN alterations that occurred along the mutation-based evolutionary trees as progressor CN alterations ., We found that arm-level gains tended to occur in founder CN alterations , while focal deletions tended to occur as progressor CN alterations ( Fig 3F ) ., To examine epigenetic ITH , we obtained DNA methylation array data for eight cases ., Cluster analysis of the multiregional methylation profiles revealed tight clustering of each case , indicating that intertumor heterogeneity ( i . e . , heterogeneity among cases ) is dominant over ITH ( Fig 4A ) ., However , we did observe substantial ITH for each case and , to analyze epigenetic ITH , we focused on variance in methylation levels of each probe in multiregional methylation profiles ., Note that the total variance can be decomposed into variance among cases and within cases ( hereafter referred as to inter- and intratumor variance , respectively ) ., Notably , we found that different categories of methylation probes contribute differently to the two types of variance ( Fig 4B , S10 Fig ) ., We categorized probes based on positional information: relative distances to CpG islands and whether the probes are located in promoter regions ., We found that probes within CpG islands tended to show higher intertumor variance than those outside CpG islands ., This is consistent with the fact that CpG island hypermethylation marks epigenetic subtypes in colorectal cancer 12; in our study , cases 5 and 9 appeared to fall into the CpG island methylator phenotype ( CIMP ) subtype as shown by cluster analysis combined with TCGA samples ( S11 Fig ) ., On the other hand , methylation probes outside CpG islands tended to show higher intratumor variance than those within CpG islands ., This observation possibly reflects the unstable nature of methylations outside CpG islands , and indicates that methylation alterations outside CpG islands are a main contributor to epigenetic ITH ., We also performed similar analysis using classification of chromosomal regions based epigenetic status in normal colon tissue ( S12 Fig ) 13 ., We found that the bivalent domains , which are marked by the H3 lysine 4 and H3 lysine 27 methylation and reportedly involved in cellar differentiation , were associated with high intertumor variance ., On the other hand , no clear association existed between any specific region category and intratumor variance , suggesting no functionality of epigenetic ITH ., To further investigate ITH and evolution in the colorectal cancer epigenomes , we identified founder and progressor methylations , as done for genetic alterations ( S13 Fig ) ., We also differentiated hyper- and hypomethylations by reference to methylation levels in paired normal mucosa samples ., Founder methylations are defined as hyper- or hypomethylations that commonly occur across all samples within each case ., As expected , we found that the loss of epigenetic gatekeepers such as SFRPs , GATA4 , and GATA5 14 is incurred by founder hypermethylation in many cases ( S14 Fig ) ., We also identified hyper- or hypomethylations that occurred along the mutation-based evolutionary trees as progressor methylations ., We then deduced temporal signatures from the multiregional methylation data by combining the three types of categorizations: founder and progressor methylation; hyper- and hypomethylation; and positional categories of methylation probes ( Fig 4C ) ., Our data showed that hypermethylations in CpG islands were more prominent in founder methylations than in progressor methylations , suggesting that CpG island hypermethylations mainly occur in the early phase of colorectal cancer evolution ., Intriguingly , connections between epigenetic aberration and aging have been reported 15 , and our analysis of TCGA data also demonstrated that patients’ ages were correlated with the number of hypermethylated probes , but not with that of hypomethylated probes ( S15 Fig ) ., Collectively , our results suggest that CpG island hypermethylations , as early events in the evolution of colorectal cancer , result from aging , consistent with our findings regarding mutations ., On the other hand , enrichment of hypomethylations in progressor methylations suggests that global hypomethylation starts at a relatively late stage in the evolution , and shapes a substantial part of epigenetic ITH ., By combining mutation , CN , and methylation profiles , we obtained integrated views of ITH in a series of colorectal cancer samples ( Fig 1 ) ., For case 3 , we additionally obtained mRNA expression profiles ., From these integrated views , as well as from evolutionary trees , we can envision the life history of each tumor ., Here , we describe that of case 3 as an example ., In the founder phase , the parental clone accumulated founder mutations together with CN gain and loss and hyper- and hypomethylation ., The founder mutation contains driver mutations represented by mutations in APC , KRAS , and FBWX7 ., In the progressor phase , the parental clone divided into two major subclones ., One major subclone had focal MYC amplification , suggesting that this major subclone was shaped by positive natural selection ., Although not having a clear driver alteration , the other major subclone had several shared CN alterations , such as 20p amplification and 1p deletion ., Then , each of the two major subclones branched into minor subclones , while accumulating many progressor alterations ., During this process generating ITH , mutation accumulations , CN alterations , and methylation alterations appeared to occur in a correlated manner ., We also found ITH in the transcriptome; notably , the major subclone harboring MYC amplification showed upregulation of the MYC expression signature together with other signatures related to cancer malignancy ., The case 3 tumor also contains a sample from liver metastasis , and the evolutionary tree suggests that the liver metastasis occurred late in the evolution , from a polypoid-like part containing the po1 sample ., The metastatic sample of case 3 is contained by the major subclone showing a lower activity of the MYC expression signature , which is unexpected if we assume that metastasis results from the acquisition of a malignant phenotype during the evolution ., In case 2 , which also contains liver metastatic samples , the metastatic samples branched out early in the evolution ., Although we need more cases to form a general rule regarding metastasis , our data demonstrate that the multiregional approach is effective to obtain information about the manner in which metastatic clones evolve ., As described so far , our genomic analysis revealed a heterogeneous evolution of colorectal cancer ., We found that PIK3CA mutations and MYC amplification occurred in the progressor phase , suggesting that a fraction of ITH is generated by positive natural selection ., However , most of the branches in the evolutionary trees had no clear evidence of such positive natural selection , and our clonality analysis of mutations suggests that ITH exists even in each of the multiregional samples ( Fig 3D and S6 Fig ) ., To clarify the principle underlying the extensive ITH , we performed computer simulation of a branching evolutionary process ( BEP ) in cancer evolution ( S16 Fig ) 16 ., In our BEP simulation , cells proliferate while accumulating random mutations in multiple genes ., Among the genes , we assume the existence of driver genes whose mutations confer a growth advantage to cells ., In the course of tumor growth , cells having mutations in driver genes are evolutionarily selected and , depending on parameter settings , cells accumulate different combinations of mutations to reproduce ITH ., Through a parameter fitting analysis , we found that the BEP simulation can generate heterogeneous mutation profiles similar to the real experimental data , if a high mutation rate , a sufficient number and sufficient strength of driver genes are assumed ., ( S17 and S18 Figs ) ., We simulated tumor evolution with such a parameter setting and then performed multiregional sequencing of the simulated tumor in silico ., Similarly to those of our 9 cases , the simulated multiregional mutation profile harbored founder , shared , and unique mutations while the heterogeneity were well correlated with geographical positions ( Fig 5A–5C ) ., Moreover , the VAF of each type of mutation tended to decrease while proceeding from founder to shared and unique mutations ( Fig 5D ) ., Note that the VAF is equal to the cancer cell fraction in which the mutation occurs , since the simulated tumors have the haploid genome and no contamination of normal cells ., Namely , this result reproduced the local ITH within each of the multiregional samples ( Fig 3D ) ., Most importantly , our BEP simulation identified a possible evolutionary principle underlying the extensive ITH ., The simulated multiregional mutation profile demonstrated that , while founder mutations occurred in most of the driver genes , progressor mutations rarely occurred in driver genes; namely , most of the progressor mutations were neutral mutations that do not confer a growth advantage ( Fig 5E ) ., Taken together with our observation that most of the branches in the evolutionary trees lacked apparent driver alteration , our data suggest that most of the ITH is generated by neutral evolution ., It should be noted that our BEP simulation could explain an origin of intertumor heterogeneity ., For a probabilistic nature of the model , independent simulation trials even with the same parameter setting generated different multiregional mutation profiles , which remind us of the multiregional mutation profiles unique to each of the nine cases ( Fig 1 and S19 Fig ) ., In this study , our integrated multiregional analysis revealed the ITH and evolutionary history of a series of nine colorectal tumors ., In particular , by focusing on founder and progressor mutations , we identified clues for decoding the life history of the tumors ., For example , we found that founder mutations included established driver mutations such as APC , KRAS , and FBWX7 , and their counts correlated with the ages of patients , suggesting that accumulation of alterations in the early phase results from aging ., It is a well-accepted dogma that cancer results from aging 17 ., Moreover , associations between somatic mutations and aging have been studied recently ., Welch et al . 18 found that acute myeloid leukemia ( AML ) genomes accumulate mutations as a function of age; furthermore , they also confirmed age-dependent mutation accumulation in hematopoietic stem/progenitor cells ., Other recent studies report that somatic mosaicism in blood increases in an age-dependent way , and it also has a positive association with cancer risk 19 , 20 ., Although the association between somatic mutations and aging has been poorly studied in the context of solid tumors , our findings indicate that an association between somatic mutations and patients’ ages exists in colorectal cancer ., During aging , a colorectal stem/progenitor cell presumably accumulates somatic mutations , some of which could unfortunately be driver mutations that transform the normal cell to a parental clone ., This view is also consistent with a recent report that a high division rate of colorectal stem/progenitor cells well explains a high lifetime risk of colorectal cancer 21 ., Through mutational signature analysis , we also found that CpG transitions at CpG sites more frequently occur in founder mutations than in progressor mutations ., This mutational signature is related to spontaneous deamination of 5-methyl-cytosine at CpG dinucleotides and is most predominantly observed in various cancer types ., A recent pan-cancer analysis 11 and our TCGA data analysis showed that this mutational signature is positively correlated with patients’ ages , which is consistent with our finding that founder mutations marked by this signature increased with patients’ ages ., As for DNA methylation , hypermethylation in CpG islands was more prominent in founder methylation than in progressor methylation ., We also found that the number of hypermethylated probes is correlated with patients’ ages in TGCA samples ., Taken together , we speculate that CpG island hypermethylation incurred by aging also predisposes a colorectal stem/progenitor cell to tumorigenesis in collaboration with somatic mutations ., Thus , genetic and epigenetic alterations are accumulated during aging , and some of them act as driver alterations that transform the normal cell to a parental clone ., Once the parental clone is established , it undergoes branched evolution in a geographically consistent way ., In addition to ITH of mutations and CN alterations , we found that epigenetic ITH marked by global hypomethylations is prevalent ., Our integrated analysis also showed that the genetic and epigenetic ITH are correlated with each other ., In contrast to founder alterations , progressor alterations appeared not to have any known driver alterations , with the exception of a few examples such as PIK3CA mutation and MYC amplification ., There also existed no parallel evolution , which is conspicuous in clear cell renal cell carcinomas 3 ., Namely , we found little evidence that positive natural selection shaped the extensive ITH , similar to the findings of recent non-small cell lung cancer studies 4 , 5 ., Moreover , our clonality analysis of mutations suggested that subclones existed even in each of the multiregional samples ., It should be noted that such local ITH is consistent with a recent breast cancer study in which single-cell sequencing identified subclonal mutations occurring at low frequencies 22 ., In pursuit of the unknown principles generating such extensive ITH , we performed the BEP simulation ., Intriguingly , our simulation suggests that neutral evolution can shape extensive ITH as observed in our multiregional mutation profiles ., Notably , our simulation also well explained the local ITH within each of the multiregional samples ., Although a single-cell mutation profile showed that a simulated tumor actually harbored numerous subclones , snapshots of the simulated evolution suggested that “macroscopic” subclones , which can be captured by the resolution of multiregional sequencing , were generated by genetic drift in the course of the neutral evolution ( S20 Fig ) ., A possible mechanism that boosts the neutral mutations is a high mutation rate , as assumed in our simulation ., We speculate that genetic instability is incurred and the mutation rate increases before the branched evolution , which is also indicated by the temporal change of mutational signatures ., Our computational analysis also suggests that a cancer stem cell hierarchy can boost the neutral evolution 16 ., Most importantly , our view that a tumor harbors numerous neutral mutations can explain the robustness and evolvability of cancer ., A therapeutic action induces an environmental change , which would convert some of the numerous neutral mutations to driver genes that confer therapeutic resistance ., Consistent with this idea , it has recently been reported that resistance to some targeted cancer drugs may result from the outgrowth of preexisting low-frequency subclones 23 ., Collectively , this work presents a new model of colorectal cancer evolution; aging leads to the accumulation of genetic and epigenetic alterations in the early phase , while neutral evolution shapes extensive ITH in the late phase ( Fig 6 ) ., Colorectal cancer has been an attractive subject for studying cancer evolution and its evolution have been addressed from various viewpoints 24–28 ., Recently , Sottoriva et al . have also proposed that ITH is mainly shaped by neutral evolution , based on uniformly high ITH , subclonal mixing in distant sites and a power-law distribution of VAFs 10 , 29 ., Along with these works , this study is unique in that it not only unveiled the extensive ITH , but also explained the underlying principle ., We believe that our model not only provides insights into colorectal cancer pathogenesis , but also constitute a new basis for designing therapeutic strategies ., Nine patients who provided written informed consent were enrolled in this study ., Detailed information about participants is provided in S1 Table ., The study protocol was reviewed and approved by Kyushu University and cooperative institutes ., All samples were obtained during surgery from patients who underwent radical resection of primary and/or liver metastatic tumors at Kyushu University Beppu Hospital , Kyushu University Hospital and Osaka University Hospital ., DNA and RNA were extracted from fresh frozen multiregional tumor samples and adjacent normal intestinal mucosa with AllPrep DNA/RNA Mini Kit ( Qiagen , Hiden , Germany ) ., For case 1–3 , high-purity tumor samples were obtained using a laser microdissection system ( Leica Laser Microdissection System; Leica Microsystems , Wetzlar , Germany ) ., A detailed protocol of sample preparation was described previously 30 ., The study design was approved by the institutional review boards and ethics committees of hospitals that made the practice of patient ( Kyushu University Hospital Institutional Review Board: Protocol Number 486–01 , Osaka University Institutional Review Board: Protocol Number 12–27 ) ., The study was conducted according to the principles expressed in the Declaration of Helsinki ., We obtained written informed consent from all the parents in this study ., There was no animal experiment in the study ., Whole exome capture was performed with The SureSelct Human All Exon 50Mb kit ( Agilent technologies ) was used for all samples ., The captured targets were subjected to massive sequencing using HiSeq 2000 ( Illumina , San Diego , CA , USA ) with the pair end 75–100 bp read option ., Information of read depth is provided in S1 Fig and S2 Table ., The sequence data were processed through an in-house pipeline ( http://genomon . hgc . jp/exome/ ) ., The sequencing reads were aligned to the NCBI Human Reference Genome Build 37 hg19 with BWA version 0 . 5 . 10 using default parameters ( http://bio-bwa . sourceforge . net/ ) ., PCR duplicate reads were removed with Picard ( http://www . picard . sourceforge . net ) ., Mutation calling was performed using the EBcall algorithm 31 with following parameters: For each case , the variants that are called in any samples and whose positions have read depths ≥ 10 in all the samples were obtained and their VAFs in all the samples were calculated by mpileup of samtools-0 . 1 . 18 ., Then , variants whose VAF > 0 . 05 were finally obtained as somatic mutations ., This step rescued mutations that were presumably shared among samples , but missed by EBcall due to disagreement with the above parameter conditions , and also filtered out variants that potentially have false negatives in any samples due to low coverage ., This procedure was applied for each case to obtain a multiregional mutation profile , from which we identified mutations shared by all the samples as founder mutations , and others as progressor mutations ., Progressor mutations were further divided into shared mutations , which were shared by a subset of samples , and unique mutations , which were unique to a single sample ., The mutations were annotated by ANNOVAR ( http://www . openbioinformatics . org/annovar/ ) ., Information of reported driver genes was based on the TCGA colon and rectum adenocarcinoma ( COADREAD ) study 8 ., Information of all the mutations is provided in S3 Table ., The multiregional mutation profile obtained for each case is visualized as a heat map whose intensities represent VAFs ., In the heat map , founder mutations were ordered along chromosomal positions , shared mutations were ordered by a hierarchical clustering , and unique mutations were sorted for samples and VAFs ., From multiregional mutation profiles , maximum parsimony trees were constructed using the maximum likelihood algorithm in the MEGA6 package 9 ., From the multiregional mutation profile of each case , we also deduced a color-coding scheme to prepare color labels of samples ., The multiregional mutation profile were regarded as an n × m matrix , whose n columns and m rows indexed n mutational positions and m samples , respectively ., We applied principle component analysis to the multiregional mutation profile and obtained the first , second and third loading vectors ., By multiplying these loading vectors , n-dimensional vectors representing mutational profiles of each sample were reduced into three-dimensional vectors ., RGB colors used for sample labels are finally papered by mixing red , green and blue proportionally to the three vector elements ., In a color-coding scheme deduced by this approach , color similarity reflects similarity of mutation profiles between samples ., We validated WES-derived mutations by targeted deep sequencing ., Preamplified cDNA library prepared for WES were captured by a custom-designed SureSelect bait library , which included: Enriched targets were sequenced and Sequencing reads were aligned to the NCBI Human Reference Genome Build 37 as done for WES ., After the reads that had either mapping quality of <25 , base quality of < 30 , or ≥ 5 mismatched bases were excluded , mutation calling was performed using following criteria: Results of the targeted deep sequencing are provided in S3 Table ., DNA was processed and hybridized to the HumanOmniExpress BeadChip Kit ( Illumina ) ., Illuminas GenomeStudio software was used to obtain B allele frequencies ( BAF ) and log R ratios ( LRR ) from the raw output data ., BAF and LRR were input into the ASCAT algorithm 32 to estimate purity and allele-specific absolute CN , which are used for calculation of CCF ., Segmented LRR was also obtained from ASCAT and used for subsequent analyses after the median was shift to, 0 . To obtain founder and progressor CN alterations , we focused on chromosomal regions subjected to arm-level and focal alterations recurrent among patients , which were reported by the TCGA study 8 ., For all the samples in each case , we obtained LRR averaged along each of the chromosomal regions ., We assumed that chromosomal regions subjected to founder CN alterations have |LRR| > 0 . 12 at least in one sample and |LRR| > 0 . 06 in all the samples ., To identify progressor CN alterations , we searched for differentially altered chromosomal regions among every pair of sample groups divided by mutation-based evolutionary tree ., The groups were prepared by focusing on branching points in the tree ., Note that , except for the first branching point that joins the trunk and branches , each branching point divided samples into two groups: those branching out from the point and the others ., For each of the chromosomal regions , we obtained difference of mean LRR between every pair of such groups , and the maximum difference as a statistic , ΔLRR ., We then obtained chromosomal regions whose |ΔLRR| > 0 . 06 as those subjected to progressor CN alterations ., For founder and progressor CN alterations identified in th | Introduction, Results, Discussion, Materials and Methods | Understanding intratumor heterogeneity is clinically important because it could cause therapeutic failure by fostering evolutionary adaptation ., To this end , we profiled the genome and epigenome in multiple regions within each of nine colorectal tumors ., Extensive intertumor heterogeneity is observed , from which we inferred the evolutionary history of the tumors ., First , clonally shared alterations appeared , in which C>T transitions at CpG site and CpG island hypermethylation were relatively enriched ., Correlation between mutation counts and patients’ ages suggests that the early-acquired alterations resulted from aging ., In the late phase , a parental clone was branched into numerous subclones ., Known driver alterations were observed frequently in the early-acquired alterations , but rarely in the late-acquired alterations ., Consistently , our computational simulation of the branching evolution suggests that extensive intratumor heterogeneity could be generated by neutral evolution ., Collectively , we propose a new model of colorectal cancer evolution , which is useful for understanding and confronting this heterogeneous disease . | Cancer is heterogeneous disease; each tumor in different patients has different cancer genomes ., Furthermore , another level of heterogeneity exists: even a single tumor harbors multiple genetically distinct subclones ., This intratumor heterogeneity is presumably one of causes of therapeutic difficulty , and its understanding is clinically necessary ., In this study , we investigated intratumor heterogeneity in colorectal cancer by analyzing sample obtained from geographically separated regions of 9 colorectal tumors ., Our integrated data analyses combined with computational simulation strongly suggest that , after clonally shared alterations were accumulated by aging , numerous subclones were generated by neutral evolution ., Importantly , this view can explain the robustness and evolvability of cancer: therapeutic action inducing an environmental change would convert some of the numerous neutral mutations to driver genes that confer therapeutic resistance ., We believe that this study not only provides insights into colorectal cancer pathogenesis , but also constitutes a new basis for designing therapeutic strategies . | medicine and health sciences, genome evolution, cancer risk factors, cancers and neoplasms, cloning, oncology, mutation, epigenetics, dna, molecular biology techniques, dna methylation, chromatin, research and analysis methods, genomics, chromosome biology, gene expression, molecular evolution, chromatin modification, dna modification, evolutionary genetics, molecular biology, biochemistry, colorectal cancer, cell biology, nucleic acids, genetics, biology and life sciences, computational biology, evolutionary biology, aging and cancer | null |
journal.pcbi.1005545 | 2,017 | Low-dimensional spike rate models derived from networks of adaptive integrate-and-fire neurons: Comparison and implementation | There is prominent evidence that information in the brain , about a particular stimulus for example , is contained in the collective neuronal spiking activity averaged over populations of neurons with similar properties ( population spike rate code ) 1 , 2 ., Although these populations can comprise a large number of neurons 3 , they often exhibit low-dimensional collective spiking dynamics 4 that can be measured using neural mass signals such as the local field potential or electroencephalography ., The behavior of cortical networks at that level is often studied computationally by employing simulations of multiple ( realistically large or subsampled ) populations of synaptically coupled individual spiking model neurons ., A popular choice of single cell description for this purpose are two-variable integrate-and-fire models 5 , 6 which describe the evolution of the fast ( somatic ) membrane voltage and an adaptation variable that represents a slowly-decaying potassium current ., These models are computationally efficient and can be successfully calibrated using electrophysiological recordings of real cortical neurons and standard stimulation protocols 5 , 7–10 to accurately reproduce their subthreshold and spiking activity ., The choice of such ( simple ) neuron models , however , does not imply reasonable ( short enough ) simulation durations for a recurrent network , especially when large numbers of neurons and synaptic connections between them are considered ., A fast and mathematically tractable alternative to simulations of large networks are population activity models in terms of low-dimensional ordinary differential equations ( i . e . , which consist of only a few variables ) that typically describe the evolution of the spike rate ., These reduced models can be rapidly solved and allow for convenient analyses of the dynamical network states using well-known methods that are simple to implement ., A popular example are the Wilson-Cowan equations 11 , which were also extended to account for ( slow ) neuronal adaptation 12 and short-term synaptic depression 13 ., Models of this type have been successfully applied to qualitatively characterize the possible dynamical states of coupled neuronal populations using phase space analyses 11–13 , yet a direct link to more biophysically described networks of ( calibrated ) spiking neurons in terms of model parameters is missing ., Recently , derived population activity models have been proposed that bridge the gap between single neuron properties and mesoscopic network dynamics ., These models are described by integral equations 14 , 15 or partial differential equations 16 , 17 Here we derive simple models in terms of low-dimensional ordinary differential equations ( ODEs ) for the spike rate dynamics of sparsely coupled adaptive nonlinear integrate-and-fire neurons that are exposed to noisy synaptic input ., The derivations are based on a Fokker-Planck equation that describes the neuronal population activity in the mean-field limit of large networks ., We develop reduced models using recent methodological advances on two different approaches: the first is based on a spectral decomposition of the Fokker-Planck operator under two different slowness assumptions 18–20 ., In the second approach we consider a cascade of linear temporal filters and a nonlinear function which are determined from the Fokker-Planck equation and semi-analytically approximated , building upon 21 ., Both approaches are extended for an adaptation current , a nonlinear spike generating current and recurrent coupling with distributed synaptic delays ., We evaluate the developed low-dimensional spike rate models quantitatively in terms of reproduction accuracy in a systematic manner over a wide range of biologically plausible parameter values ., In addition , we provide numerical implementations for the different reduction methods as well as the Fokker-Planck equation under a free license as open source project ., For the derived models in this contribution we use the adaptive exponential integrate-and-fire ( aEIF ) model 5 to describe individual neurons , which is similar to the model proposed by Izhikevich 6 but includes biophysically meaningful parameters and a refined description of spike initiation ., However , the presented derivations are equally applicable when using the Izhikevich model instead ( requiring only a small number of simple substitutions in the code ) ., Through their parameters the derived models retain a direct , quantitative link to the underlying spiking model neurons , and they are described in a well-established , convenient form ( ODEs ) that can be rapidly solved and analyzed ., Therefore , these models are well suited, ( i ) for mathematical analyses of dynamical states at the population level , e . g . , linear stability analyses of attractors , and, ( ii ) for application in multi-population brain network models ., Apart from a specific network setting , the derived models are also appropriate as a spike rate description of individual neurons under noisy input conditions ., The structure of this article contains mildly redundant model specifications allowing the readers who are not interested in the methodological foundation to directly read the self-contained Sect ., Results ., The quantity of our interest is the population-averaged number of spikes emitted by a large homogeneous network of N sparsely coupled aEIF model neurons per small time interval , i . e . , the spike rate rN ( t ) ., The state of neuron i at time t is described by the membrane voltage Vi ( t ) and adaptation current wi ( t ) , which evolve piecewise continuously in response to overall synaptic current Isyn , i = Iext , i ( t ) + Irec , i ( t ) ., This input current consists of fluctuating network-external drive Iext , i = Cμext ( t ) + σext ( t ) ξext , i ( t ) with membrane capacitance C , time-varying moments μext , σ ext 2 and unit Gaussian white noise process ξext , i as well as recurrent input Irec , i ., The latter causes delayed postsynaptic potentials ( i . e . , deflections of Vi ) of small amplitude J triggered by the spikes of K presynaptic neurons ( see Sect . Methods for details ) ., Here we present two approaches of how the spike rate dynamics of the large , stochastic delay-differential equation system for the 2N states ( Vi , wi ) can be described by simple models in terms of low-dimensional ODEs ., Both approaches, ( i ) take into account adaptation current dynamics that are sufficiently slow , allowing to replace the individual adaptation current wi by its population-average 〈w〉 , governed by, d ⟨ w ⟩ d t = a ( ⟨ V ⟩ ∞ - E w ) - ⟨ w ⟩ τ w + b r ( t ) , ( 1 ), where a , Ew , b , τw are the adaptation current model parameters ( subthreshold conductance , reversal potential , spike-triggered increment , time constant , respectively ) , 〈V〉∞ is the steady-state membrane voltage averaged across the population ( which can vary over time , see below ) , and r is the spike rate of the respective low-dimensional model ., Furthermore , both approaches, ( ii ) are based on the observation that the collective dynamics of a large , sparsely coupled ( and noise driven ) network of integrate-and-fire type neurons can be well described by a Fokker-Planck equation ., In this intermediate Fokker-Planck ( FP ) model the overall synaptic input is approximated by a mean part with additive white Gaussian fluctuations , Isyn , i/C ≈ μsyn ( t , rd ) + σsyn ( t , rd ) ξi ( t ) , that are uncorrelated between neurons ., The moments of the overall synaptic input ,, μ syn = μ ext ( t ) + J K r d ( t ) , σ syn 2 = σ ext 2 ( t ) + J 2 K r d ( t ) , ( 2 ), depend on time via the moments of the external input and , due to recurrent coupling , on the delayed spike rate rd ., The latter is governed by, d r d d t = r - r d τ d , ( 3 ), which corresponds to individual propagation delays drawn from an exponentially distributed random variable with mean τd ., The FP model involves solving a partial differential equation ( PDE ) to obtain the time-varying membrane voltage distribution p ( V , t ) and the spike rate r ( t ) ., The first reduction approach is based on the spectral decomposition of the Fokker-Planck operator L and leads to the following two low-dimensional models: the “basic” model variant ( spec1 ) is given by a complex-valued differential equation describing the spike rate evolution in its real part ,, d r ˜ d t = λ 1 ( r ˜ - r ∞ ) , r ( t ) = Re { r ˜ } , ( 4 ), where λ1 ( μtot , σtot ) is the dominant eigenvalue of L and r∞ ( μtot , σtot ) is the steady-state spike rate ., Its parameters λ1 , r∞ , and 〈V〉∞ ( cf ., Eq ( 1 ) ) depend on the total input moments given by μtot ( t ) = μsyn − 〈w〉/C and σ tot 2 ( t ) = σ syn 2 which closes the model ( Eqs ( 1 ) – ( 4 ) ) ., The other , “advanced” spectral model variant ( spec2 ) is given by a real-valued second order differential equation for the spike rate ,, β 2 r ¨ + β 1 r ˙ + β 0 r = r ∞ - r - β c , ( 5 ), where the dots denote time derivatives ., Its parameters β2 , β1 , β0 , βc , r∞ and 〈V〉∞ depend on the total input moments ( μtot , σ tot 2 ) as follows: the latter two parameters explicitly as in the basic model above , the former four indirectly via the first two dominant eigenvalues λ1 , λ2 and via additional quantities obtained from the ( stationary and the first two nonstationary ) eigenfunctions of L and its adjoint L * ., Furthermore , the parameter βc depends explicitly on the population-averaged adaptation current 〈w〉 , the delayed spike rate rd , and on the first and second order time derivatives of the external moments μext and σ ext 2 . The second approach is based on a Linear-Nonlinear ( LN ) cascade , in which the population spike rate is generated by applying to the time-varying mean and standard deviation of the overall synaptic input , μsyn and σsyn , separately a linear temporal filter , followed by a common nonlinear function ., These three components–two linear filters and a nonlinearity–are extracted from the Fokker-Planck equation ., Approximating the linear filters using exponentials and damped oscillating functions yields two model variants: In the basic “exponential” ( LNexp ) model the filtered mean μf and standard deviation σf of the overall synaptic input are given by, d μ f d t = μ syn - μ f τ μ , d σ f d t = σ syn - σ f τ σ , ( 6 ), where the time constants τμ ( μeff , σeff ) , τσ ( μeff , σeff ) depend on the effective ( filtered ) input mean μeff ( t ) = μf − 〈w〉/C and standard deviation σeff ( t ) = σf ., The “damped oscillator” ( LNdos ) model variant , on the other hand , describes the filtered input moments by, μ ¨ f + 2 τ μ ˙ f + ( 2 τ 2 + ω 2 ) μ f = 1 + τ 2 ω 2 τ ( μ syn τ + μ ˙ syn ) , ( 7 ) d σ f d t = σ syn - σ f τ σ , ( 8 ), where the time constants τ ( μtot , σtot ) , τσ ( μtot , σtot ) and the angular frequency ω ( μtot , σtot ) depend on the total input moments defined above ., In both LN model variants the spike rate is obtained by the nonlinear transformation of the effective input moments through the steady-state spike rate ,, r ( t ) = r ∞ ( μ eff , σ eff ) , ( 9 ), and the steady-state mean membrane voltage 〈V〉∞ ( cf ., Eq ( 1 ) ) is also evaluated at ( μeff , σeff ) ., These four models ( spec1 , spec2 , LNexp , LNdos ) from both reduction approaches involve a number of parameters that depend on the strengths of synaptic input and adaptation current only via the total or effective input moments ., We refer to these parameters as quantities below to distinguish them from fixed ( independent ) parameters ., The computational complexity when numerically solving the models forward in time ( for different parametrizations ) can be greatly reduced by precomputing those quantities for a range of values for the total/effective input moments and using look-up tables during time integration ., Changing any parameter value of the external input , the recurrent coupling or the adaptation current does not require renewed precomputations , enabling rapid explorations of parameter space and efficient ( linear ) stability analyses of network states ., The full specification of the “ground truth” system ( network of aEIF neurons ) , the derivations of the intermediate description ( FP model ) and the low-dimensional spike rate models complemented by concrete numerical implementations are provided in Sect ., Methods ( that is complemented by the supporting material S1 Text ) ., In Fig 1 we visualize the outputs of the different models using an example excitatory aEIF network exposed to external input with varying mean μext ( t ) and standard deviation σext ( t ) ., Here , and in the subsequent two sections , we assess the accuracy of the four low-dimensional models to reproduce the spike rate dynamics of the underlying aEIF population ., The intermediate FP model is included for reference ., The derived models generate population activity in response to overall synaptic input moments μsyn and σ syn 2 . These depend on time via the external moments μext ( t ) and σ ext 2 ( t ) , and the delayed spike rate rd ( t ) ., Therefore , it is instrumental to first consider an uncoupled population and suitable variations of external input moments that effectively mimic a range of biologically plausible presynaptic spike rate dynamics ., This allows us to systematically compare the reproduction performance of the different models over a manageable parameter space ( without K , J , τd ) , yet it provides useful information on the accuracy for recurrent networks ., For many network settings the dominant effect of synaptic coupling is on the mean input ( cf ., Eq ( 2 ) ) ., Therefore , we consider first in detail time-varying mean but constant variance of the input ., Specifically , to account for a wide range of oscillation frequencies for presynaptic spike rates , μext is described by an Ornstein-Uhlenbeck ( OU ) process, μ ˙ ext = μ ¯ - μ ext τ ou μ + 2 τ ou μ ϑ μ ξ ( t ) , ( 10 ), where τ ou μ denotes the correlation time , μ ¯ and ϑμ are the mean and standard deviation of the stationary normal distribution , i . e . , lim t → ∞ μ ext ( t ) ∼ N ( μ ¯ , ϑ μ 2 ) , and ξ is a unit Gaussian white noise process ., Sample time series generated from the OU process are filtered using a Gaussian kernel with a small standard deviation σt to obtain sufficiently differentiable time series μ ˜ ext ( due to the requirements of the spec2 model and the LNdos model ) ., The filtered realization μ ˜ ext ( t ) is then used for all models to allow for a quantitative comparison of the different spike rate responses to the same input ., The value of σt we use in this study effectively removes very large oscillation frequencies which are rarely observed , while lower frequencies 22 are passed ., The parameter space we explore covers large and small correlation times τ ou μ , strong and weak input mean μ ¯ and standard deviation σext , and for each of these combinations we consider an interval from small to large variation magnitudes ϑμ ., The values of τ ou μ and ϑμ determine how rapid and intense μext ( t ) fluctuates ., We apply two performance measures , as in 21 ., One is given by Pearson’s correlation coefficient ,, ρ ( r N , r ) ≔ ∑ k = 1 M ( r N ( t k ) - r ¯ N ) ( r ( t k ) - r ¯ ) ∑ k = 1 M ( r N ( t k ) - r ¯ N ) 2 ∑ k = 1 M ( r ( t k ) - r ¯ ) 2 , ( 11 ), between the ( discretely given ) spike rates of the aEIF population and each derived model with time averages r ¯ N = 1 / M ∑ k = 1 M r N ( t k ) and r ¯ = 1 / M ∑ k = 1 M r ( t k ) over a time interval of length tM − t1 ., For comparison , we also include the correlation coefficient between the aEIF population spike rate and the time-varying mean input , ρ ( rN , μext ) ., In addition , to assess absolute differences we calculate the root mean square ( RMS ) distance ,, dRMS ( rN , r ) ≔1M∑k=1M ( rN ( tk ) −r ( tk ) ) 2 , ( 12 ), where M denotes the number of elements of the respective time series ( rN , r ) ., We find that three of the four low-dimensional spike rate models ( spec2 , LNexp , LNdos ) very well reproduce the spike rate rN of the aEIF neurons: for the LNexp model ρ > 0 . 95 and for the spec2 and LNdos models ρ ≳ 0 . 8 ( each ) over the explored parameter space , see Fig 2 . Only the basic spectral model ( spec1 ) is substantially less accurate ., Among the best models , the simplest ( LNexp ) overall outperforms spec2 and LNdos , in particular for fast and strong mean input variations ., However , in the strongly mean-driven regime the best performing model is spec2 ., We observe that the performance of any of the spike rate models decreases ( with model-specific slope ) with, ( i ) increasing variation strength ϑμ larger than a certain ( small ) value , and with, ( ii ) smaller τ ou μ , i . e . , faster changes of μext ., For small values of ϑμ fluctuations of rN , which are caused by the finite aEIF population size N and do not depend on the fluctuations of μext , deteriorate the performance measured by ρ ( see also 21 , p . 13 right ) ., This explains why ρ does not increase as ϑμ decreases ( towards zero ) for any of the models ., Naturally , the FP model is the by far most accurate spike rate description in terms of both measures , correlation coefficient ρ and RMS distance ., This is not surprising because the four low-dimensional models are derived from that ( infinite-dimensional ) representation ., Thus , the performance of the FP system defines an upper bound on the correlation coefficient ρ and a lower bound on the RMS distance for the low-dimensional models ., In detail: for moderately fast changing mean input ( large τ ou μ ) the three models spec2 , LNexp and LNdos exhibit excellent reproduction performance with ρ > 0 . 95 , and spec1 shows correlation coefficients of at least ρ = 0 . 9 ( Fig 2A ) , which is substantially better than ρ ( rN , μext ) ., The small differences between the three top models can be better assessed from the RMS distance measure ., For large input variance σ ext 2 the two LN models perform best ( cf . Fig 2A , top , and for an example , 2C ) ., For weak input variance and large mean ( small σext , large μ ¯ ) the spec2 model outperforms the LN models , unless the variation magnitude ϑμ is very large ., For small mean μ ¯ , where transient activity is interleaved with periods of quiescence , the LNexp model performs best , except for weak variations ϑμ , where LNdos is slightly better ( see Fig 2A , bottom ) ., Stronger differences in performance emerge when considering faster changes of the mean input μext ( t ) ( i . e . , for small τ ou μ ) , see Fig 2B , and for examples , Fig 2C ., The spec1 model again performs worst with ρ values even below the input/output correlation baseline ρ ( rN , μext ) for large mean input μ ¯ ( cf . Fig 2B , left ) ., The spec1 spike rate typically decays too slowly ( cf . Fig 2C ) ., The three better performing models differ as follows: for large input variance and mean ( large σext and μ ¯ ) , where the spike rate response to the input is rather fast ( cf ., increased ρ ( rN , μext ) ) , the performance of all three models in terms of ρ is very high , but the RMS distance measure indicates that LNexp is the most accurate model ( cf . Fig 2B , top ) ., For weak mean input LNexp is once again the top model while LNdos and , especially noticeable , spec2 show a performance decline ( see example in Fig 2C ) ., For weak input variance ( Fig 2A , bottom ) , where significant ( oscillatory ) excursions of the spike rates in response to changes in the mean input can be observed ( see also Fig 1 ) , we obtain the following benchmark contrast: for large mean drive μ ¯ the spec2 model performs best , except for large variation amplitudes ϑμ , at which LNexp is more accurate ., Smaller mean input on the other hand corresponds to the most sensitive regime where periods of quiescence alternate with rapidly increasing and decaying spike rates ., The LNexp model shows the most robust and accurate spike rate reproduction in this setting , while LNdos and spec2 each exhibit decreased correlation and larger RMS distances–spec2 even for moderate input variation intensities ϑμ ., The slowness approximation underlying the spec2 model likely induces an error due to the fast external input changes in comparison with the rather slow intrinsic time scale by the dominant eigenvalue , τ ou μ = 5 ms vs . 1/|Re{λ1}| ≈ 15 ms ( cf . visualization of the spectrum in Sect . Spectral models ) ., Note that for these weak inputs the distribution of the spike rate is rather asymmetric ( cf . Fig 2B ) ., Interestingly the LNdos model performs worse than LNexp for large mean input variations ( i . e . , large ϑμ ) in general , and only slightly better for small input variance and mean input variations that are not too large and fast ., We would like to note that decreasing the Gaussian filter width σt to smaller values , e . g . , fractions of a millisecond , can lead to a strong performance decline for the spec2 model because of its explicit dependence on first and second order time derivatives of the mean input ., Furthermore , we show how the adaptation parameters affect the reproduction performance of the different models in Fig 3 . The adaptation time constant τw and spike-triggered adaptation increment b are varied simultaneously ( keeping their product constant ) such that the average spike rate and adaptation current , and thus the spiking regime , remain comparable for all parametrizations ., As expected , the accuracy of the derived models decreases for faster adaptation current dynamics , due to the adiabatic approximation that relies on sufficiently slow adaptation ( cf . Sect . Methods ) ., Interestingly however , the performance of all reduced models ( except spec1 ) declines only slightly as the adaptation time constant decreases to the value of the membrane time constant ( which means the assumption of separated time scales underlying the adiabatic approximation is clearly violated ) ., This kind of robustness is particularly pronounced for input with large baseline mean μ ¯ and small noise amplitude σext , cf ., Fig 3B ., For perfectly balanced excitatory and inhibitory synaptic coupling the contribution of presynaptic activity to the mean input μsyn is zero by definition , but the input variance σ syn 2 is always positively ( linearly ) affected by a presynaptic spike rate–even for a negative synaptic efficacy J ( cf ., Eq ( 2 ) ) ., To assess the performance of the derived models in this scenario , but within the reference setting of an uncoupled population , we consider constant external mean drive μext and let the variance σ ext 2 ( t ) evolve according to a filtered OU process ( such as that used for the mean input μext in the previous section ) with parameters σ 2 ¯ and ϑσ2 of the stationary normal distribution N ( σ 2 ¯ , ϑ σ 2 2 ) , correlation time τ ou σ 2 and Gaussian filter standard deviation σt as before ., The results of two input parametrizations are shown in Fig 4 . For large input mean μext and rapidly varying variance σ ext 2 ( t ) the spike rate response of the aEIF population is very well reproduced by the FP model and , to a large extent , by the spec2 model ( cf . Fig 4A ) ., This may be attributed to the fact that the latter model depends on the first two time derivatives of the input variance σ ext 2 . The LN models cannot well reproduce the rapid spike rate excursions in this setting , and the spec1 model performs worst , exhibiting time-lagged spike rate dynamics compared to rN ( t ) which leads to a very small value of correlation coefficient ρ ( below the input/output correlation baseline ρ ( r N , σ ext 2 ) ) ., For smaller mean input μext and moderately fast varying variance σ ext 2 ( t ) ( larger correlation time τ ou σ 2 ) the fluctuating aEIF population spike rate is again nicely reproduced by the FP model while the rate response of the spec2 model exhibits over-sensitive behavior to changes in the input variance , as indicated by the large RMS distance ( see Fig 4B ) ., This effect is even stronger for faster variations , i . e . , smaller τ ou σ 2 ( cf . supplementary visualization S1 Fig ) ., The LN models perform better in this setting , and the spec1 model ( again ) performs worst in terms of correlation coefficient ρ due to its time-lagged spike rate response ., It should be noted that the lowest possible value of the input standard deviation , i . e . , σext ( plus a nonnegative number in case of recurrent input ) cannot be chosen completely freely but must be large enough ( ≳ 0 . 5 mV / ms ) for our parametrization ., This is due to theoretical reasons ( Fokker-Planck formalism ) and practical reasons ( numerics for Fokker-Planck solution and for calculation of the derived quantities , such as r∞ ) ., To demonstrate the applicability of the low-dimensional models for network analyses we consider a recurrently coupled population of aEIF neurons that produces self-sustained network oscillations by the interplay of strong excitatory feedback and spike-triggered adaptation or , alternatively , by delayed recurrent synaptic inhibition 16 , 23 ., The former oscillation type is quite sensitive to changes in input , adaptation and especially coupling parameters for the current-based type of synaptic coupling considered here and due to lack of ( synaptic ) inhibition and refractoriness ., For example , a small increase in coupling strength can lead to a dramatic ( unphysiologic ) increase in oscillation amplitude because of strong recurrent excitation ., Hence we consider a difficult setting here to evaluate the reduced spike rate models–in particular , when the network operates close to a bifurcation ., In Fig 5A we present two example parametrizations from a region ( in parameter space ) that is characterized by stable oscillations ., This means the network exhibits oscillatory spike rate dynamics for constant external input moments μext and σ ext 2 . The derived models reproduce the limit cycle behavior of the aEIF network surprisingly well , except for small frequency and amplitude deviations ( FP , spec2 , LNdos , LNexp ) and larger frequency mismatch ( spec1 ) , see Fig 5A , top ., For weaker input moments and increased spike-triggered adaptation strength the network is closer to a Hopf bifurcation 16 , 23 ., It is , therefore , not surprising that the differences in oscillation period and amplitude are more prominent ( cf . Fig 5A , bottom ) ., The bifurcation point of the LNexp model is slightly shifted , shown by the slowly damped oscillatory convergence to a fixed point ., This suggests that the bifurcation parameter value of each of the derived models is not far from the true critical parameter value of the aEIF network but can quantitatively differ ( slightly ) in a model-dependent way ., The second type of oscillation is generated by delayed synaptic inhibition 22 and does not depend on the ( neuronal ) inhibition that is provided by an adaptation current ., To demonstrate this independence the adaptation current was disabled ( by setting the parameters a = b = 0 ) for the two respective examples that are shown in Fig 5B ., Similarly as for the previous oscillation type , the low-dimensional models ( except spec1 ) reproduce the spike rate limit cycle of the aEIF network surprisingly well , in particular for weak external input ( see Fig 5B , top ) ., For larger external input and stronger inhibition with shorter delay the network operates close to a Hopf bifurcation , leading to larger differences in oscillation amplitude and frequency in a model-dependent way ( Fig 5B , bottom ) ., Note that the intermediate ( Fokker-Planck ) model very well reproduces the inhibition-based type of oscillation which demonstrates the applicability of the underlying mean-field approximation ., We would also like to note that enabling the adaptation current dynamics ( only ) leads to decreased average spike rates but does not affect the reproduction accuracy ., We would like to emphasize that the previous comprehensive evaluations for an uncoupled population provide a deeper insight on the reproduction performance–also for a recurrent network–than the four examples shown here , as explained in the Sect ., Performance for variations of the mean input ., For example , the ( improved ) reproduction performance for increased input variance in the uncoupled setting ( cf . Fig 2 ) informs about the reproduction performance for networks of excitatory and inhibitory neurons that are roughly balanced , i . e . , where the overall input mean is rather small compared to the input standard deviation ., We have developed efficient implementations of the derived models using the Python programming language and by employing the library Numba for low-level machine acceleration 24 ., These include:, ( i ) the numerical integration of the Fokker-Planck model using an accurate finite volume scheme with implicit time discretization ( cf . Sect . Methods ) ,, ( ii ) the parallelized precalculation of the quantities required by the low-dimensional spike rate models and, ( iii ) the time integration of the latter models , as well as example scripts demonstrating, ( i ) –, ( iii ) ., The code is available as open source software under a free license at GitHub: https://github . com/neuromethods/fokker-planck-based-spike-rate-models With regards to computational cost , summarizing the results of several aEIF network parametrizations , the duration to generate population activity time series for the low-dimensional spike rate models is usually several orders of magnitude smaller compared to numerical simulation of the original aEIF network and a few orders of magnitude smaller in comparison to the numerical solution of the FP model ., For example , considering a population of 50 , 000 coupled neurons with 2% connection probability , a single simulation run of 5 s and the same integration time step across the models , the computation times amounted to 1 . 1–3 . 6 s for the low-dimensional models ( with order–fast to slow–LNexp , spec1 , LNdos , spec2 ) , about 100 s for the FP model and roughly 1500 s for the aEIF network simulation on a dual-core laptop computer ., The time difference to the network simulation substantially increases with the numbers of neurons and connections , and with spiking activity within the network due to the extensive propagation of synaptic events ., Note that the speedup becomes even more pronounced with increasing number of populations , where the runtimes of the FP model and the aEIF network simulation scale linearly and the low-dimensional models show a sublinear runtime increase due to vectorization of the state variables representing the different populations ., The derived low-dimensional ( ODE ) spike rate models are very efficient to integrate given that the required input-dependent parameters are available as precalulated look-up quantities ., For the grids used in this contribution , the precomputation time was 40 min . for the cascade ( LNexp , LNdos ) models and 120 min . for the spectral ( spec1 , spec2 ) models , both on a hexa-core desktop computer ., The longer calculation time for the spectral models was due to the finer internal grid for the mean input ( see S1 Text ) ., Note that while the time integration of the spec2 model is on the same order as for the other low-dimensional models its implementation complexity is larger because of the many quantities it depends on , cf ., Eqs ( 63 ) – ( 66 ) ., In addition to the work we build upon 18–21 ( cf . Sect . Methods ) there are a few other approaches to derive spike rate models from populations of spiking neurons ., Some methods also result in an ODE system , taking into account ( slow ) neuronal adaptation 17 , 26 , 36–38 or disregarding it 39 ., The settings differ from the work presented here in that, ( i ) the intrinsic neuronal dynamics are adiabatically neglected 17 , 26 , 36 , 37 ,, ( ii ) only uncoupled populations 38 or all-to-all connected networks 17 , 36 , 39 are assumed in contrast to sparse connectivity , and, ( iii ) ( fixed ) heterogeneous instead of fluctuating input is considered 39 ., Notably , these | Introduction, Results, Discussion, Methods | The spiking activity of single neurons can be well described by a nonlinear integrate-and-fire model that includes somatic adaptation ., When exposed to fluctuating inputs sparsely coupled populations of these model neurons exhibit stochastic collective dynamics that can be effectively characterized using the Fokker-Planck equation ., This approach , however , leads to a model with an infinite-dimensional state space and non-standard boundary conditions ., Here we derive from that description four simple models for the spike rate dynamics in terms of low-dimensional ordinary differential equations using two different reduction techniques: one uses the spectral decomposition of the Fokker-Planck operator , the other is based on a cascade of two linear filters and a nonlinearity , which are determined from the Fokker-Planck equation and semi-analytically approximated ., We evaluate the reduced models for a wide range of biologically plausible input statistics and find that both approximation approaches lead to spike rate models that accurately reproduce the spiking behavior of the underlying adaptive integrate-and-fire population ., Particularly the cascade-based models are overall most accurate and robust , especially in the sensitive region of rapidly changing input ., For the mean-driven regime , when input fluctuations are not too strong and fast , however , the best performing model is based on the spectral decomposition ., The low-dimensional models also well reproduce stable oscillatory spike rate dynamics that are generated either by recurrent synaptic excitation and neuronal adaptation or through delayed inhibitory synaptic feedback ., The computational demands of the reduced models are very low but the implementation complexity differs between the different model variants ., Therefore we have made available implementations that allow to numerically integrate the low-dimensional spike rate models as well as the Fokker-Planck partial differential equation in efficient ways for arbitrary model parametrizations as open source software ., The derived spike rate descriptions retain a direct link to the properties of single neurons , allow for convenient mathematical analyses of network states , and are well suited for application in neural mass/mean-field based brain network models . | Characterizing the dynamics of biophysically modeled , large neuronal networks usually involves extensive numerical simulations ., As an alternative to this expensive procedure we propose efficient models that describe the network activity in terms of a few ordinary differential equations ., These systems are simple to solve and allow for convenient investigations of asynchronous , oscillatory or chaotic network states because linear stability analyses and powerful related methods are readily applicable ., We build upon two research lines on which substantial efforts have been exerted in the last two decades:, ( i ) the development of single neuron models of reduced complexity that can accurately reproduce a large repertoire of observed neuronal behavior , and, ( ii ) different approaches to approximate the Fokker-Planck equation that represents the collective dynamics of large neuronal networks ., We combine these advances and extend recent approximation methods of the latter kind to obtain spike rate models that surprisingly well reproduce the macroscopic dynamics of the underlying neuronal network ., At the same time the microscopic properties are retained through the single neuron model parameters ., To enable a fast adoption we have released an efficient Python implementation as open source software under a free license . | medicine and health sciences, action potentials, neural networks, engineering and technology, signal processing, membrane potential, electrophysiology, neuroscience, signal filtering, mathematics, algebra, network analysis, computer and information sciences, animal cells, linear filters, approximation methods, cellular neuroscience, cell biology, linear algebra, physiology, neurons, biology and life sciences, cellular types, physical sciences, eigenvalues, neurophysiology | null |
journal.pntd.0001417 | 2,012 | Low Efficacy of Single-Dose Albendazole and Mebendazole against Hookworm and Effect on Concomitant Helminth Infection in Lao PDR | Infections with the three common soil-transmitted helminths ( STHs ) , Ascaris lumbricoides , Trichuris trichiura , and hookworm ( Ancylostoma duodenale and Necator americanus ) , are a global public-health concern , particularly in areas where poor sanitation prevails 1 , 2 ., STH infections are among the most widespread of the neglected tropical diseases ( NTDs ) 3 ., Indeed , more than a billion people are currently infected with one or several STH species , even though growing efforts are underway to control these parasitic worm infections 4 ., In terms of their estimated global burden , hookworm is the most important among the STHs , perhaps responsible for more than 20 million disability-adjusted life years ( DALYs ) among the estimated 600 million infected people worldwide 1 , 5 ., Chronic hookworm infection cause intestinal blood loss and result in poor iron status and iron-deficiency anemia , particularly in children , and women in reproductive age 1 , 6 , 7 ., As a consequence , permanent impairment , including delayed physical and cognitive development , has been described 8 ., In the absence of a vaccine , the global strategy to control STHs and other NTDs is to reduce morbidity through repeated large-scale administration of anthelmintic drugs , a strategy phrased preventive chemotherapy 9 ., At present , the World Health Organization ( WHO ) recommends four drugs against STH infections , of which albendazole and mebendazole are the two most widely used drugs for preventive chemotherapy 10 ., In 2008 , in the Western Pacific Region , 33 . 4 million children were given anthelmintic drugs 11 ., According to the Lao national scheme for school deworming , there is a treatment round at the beginning of the first semester ( September–December ) and in the second semester ( January–April ) ., Mebendazole ( single 500 mg oral dose ) is annually distributed to all school-aged children since 2005 12 ., Recent efforts have been made to provide mebendazole also to preschool-aged children through the Expanded Program on Immunization ( EPI ) and alongside vitamin A distribution campaigns 4 , 13 ., However , the efficacy of mebendazole and albendazole against STH infections in Lao PDR remains to be determined , and such locally derived evidence is important to guide the national treatment policy ., The liver fluke Opisthorchis viverrini is co-endemic in Lao PDR , and particularly high prevalences have been observed in the southern provinces 14–17 ., Praziquantel is the current drug of choice against O . viverrini 3 ., Previous work has shown that multiple doses of albendazole also show some effect 18 , 19 ., Hence , in areas where STHs and O . viverrini co-exist and preventive chemotherapy targeting STHs is under way , it will be interesting to monitor for potential ancillary benefits of this control strategy against opisthorchiasis ., The purpose of this study was to assess the efficacy of single-dose albendazole ( 400 mg ) and single-dose mebendazole ( 500 mg ) against hookworm infection among school-aged children in Lao PDR ., In addition , the effect on other STHs ( i . e . , A . lumbricoides and T . trichiura ) and O . viverrini in co-infected children was assessed ., Our study complements a recent investigation , done in the Peoples Republic of China that compared single and triple dosing with albendazole and mebendazole against the three common STHs 20 ., The research protocol ( see Protocol S1 ) was approved by the Ethics Committee of Basel , Switzerland ( EKBB; reference no . 146/08 ) and the Lao National Ethics Committee for Health Research ( NECHR ) , Ministry of Health , Vientiane , Lao PDR ( reference no . 170/NECHR ) ., The trial is registered with Current Controlled Trials ( identifier: ISRCTN29126001 ) ., Written informed consent was obtained from parents/legal guardians of eligible children ., Participation was voluntary and children could withdraw from the trial at any time without further obligation ., At completion of the trial , all children of the two primary schools and participants who were still found positive for hookworm ( or other STHs ) were treated with albendazole ( 400 mg ) ., O . viverrini-infected children were administered praziquantel according to national guidelines 21 ., A randomized , open-label trial was carried out between February and March 2009 in two primary schools ( Oudomsouk and Nongbok Noi ) in Batieng district , Champasack province , southern Lao PDR ., Children in the two schools were treated with mebendazole 5–6 months prior to the start of our study ., The schools are located approximately 15 km southeast of Pakse town , on the Bolaven plateau at an altitude of approximately 1 , 000 m above sea level ( geographical coordinates: 105°56′53″N latitude , 15°14′59″E longitude ) ., The rainy season lasts from May to mid-October ., A census done in 2007 revealed that 43 , 651 people lived in the 95 villages of Batieng district ( Dr . Nanthasane Vannavong , Champasack Provincial Health Department; personal communication ) ., More than three-quarter of the households ( 77 . 5% ) lack appropriate sanitation ., Drinking water is primarily obtained from unprotected boreholes and wells ., Most villagers live on subsistence rice farming and rubber plantations ( Dr . Nanthasane Vannavong , Champasak Provincial Health Department; personal communication ) ., Infections with STHs and O . viverrini are common in Batieng district; a recent study revealed infection prevalences above 50% and above 20% , respectively 22 ., We designed a randomized , open-label trial comparing albendazole ( single 400 mg dose ) and mebendazole ( single 500 mg dose ) for treatment of hookworm infection ., The sample size was calculated based on results of a meta-analysis on the efficacy of current anthelmintic drugs against common STH infections , which reported cure rates ( CR; defined as percentage of helminth-positive individuals who became helminth-egg negative after treatment ) of 75% and 15% for albendazole ( 400 mg ) and mebendazole ( 500 mg ) , respectively against hookworm infection 10 ., In order to account for the large variation ( uncertainty ) of the observed efficacy of mebendazole in the individual studies ( CRs of 8–91% were found in the six randomized controlled trials ) , we more than tripled the mean efficacy of mebendazole ( 50% instead of 15% ) ., Assuming superiority of albendazole ( 1-tailed test ) and taking into account a 90% power , and an alpha error of 5% , we obtained a sample size of 85 children per treatment group ., Furthermore , we assumed a drop-out rate of 15% , which resulted in a total sample size of 200 hookworm-positive school-aged children ., The teachers of the two primary schools , the children , and the staff of the National Institute of Public Health , Centre of Malaria , Parasitology and Entomology , Centre for Laboratory and Epidemiology , the Provincial Department of Health , the Provincial Hospital , and the Malaria Station of Champassak , and the village authorities were informed one week in advance on the study aims and procedures ., Potential risks and benefits were explained to all children and their parents/guardians ., An informed consent form was distributed to all parents/guardians and signed ., Children assented orally ., At baseline screening the consenting children ( n\u200a=\u200a465 ) of the two schools , aged 6–12 years , provided two fresh stool samples within a period of 3 days ., Stool containers were filled by children and transferred to a laboratory in the early morning ( between 8 and 9 am ) ., All collected specimens were worked up on the day of collection ., From each stool sample , duplicate Kato-Katz thick smears were prepared on microscope slides , using standard 41 . 7 mg templates 23 ., Kato-Katz thick smears were quantitatively examined under a light microscope for helminths with a 100× magnification ., Slides were read within 30–45 min after preparation ., A random sample of approximately 10% of the Kato-Katz thick smears were re-examined by two senior technicians for quality control purposes ., In case of discrepancies ( i . e . , positive vs . negative results and egg counts differing by >10% ) , results were discussed with the respective technicians , and the slides re-examined until agreement was reached ., In addition , a questionnaire was administered to each participating child to obtain sociodemographic data ( i . e . , sex , age , parents education and occupation , ethnic group , and sanitation infrastructure ) , and behavioral data ( i . e . , wearing shoes , sources of drinking water , food consumption , and personal hygiene ) ., Hookworm-positive children ( defined by the presence of at least one hookworm egg in one of the quadruplicate Kato-Katz thick smears examined per child ) were invited for treatment ( n\u200a=\u200a200 ) ., At enrollment , a clinical examination , which included measurement of weight ( using an electronic balance measured to the nearest 0 . 1 kg ) , height ( using a measuring tap fixed to the wall and measured to the nearest cm ) , and axcillary temperature ( using battery-powered thermometers , measured to the nearest 0 . 01°C ) , anemia assessment ( finger prick capillary blood sample ) was conducted , and a medical history taken ., Children were excluded if they had fever , or showed signs of severe malnutrition ., Additional exclusion criteria were the presence of any abnormal medical condition such as cardiac , vascular , pulmonary , gastrointestinal , endocrine , neurologic , hematologic ( e . g . , thalassaemia ) , rheumatologic , psychiatric , or metabolic disturbances , recent history of anthelmintic treatment ( e . g . , albendazole , mebendazole , pyrantel pamoate , levamisole , ivermectin , and praziquantel ) , attending other clinical trials during the study , or reported hypersensitivity to albendazole or mebendazole ., At follow-up , 21–23 days after drug administration , two stool samples were collected from each child and transferred to a hospital laboratory within one hour after collection ., Each stool specimen collected at follow-up was subjected to the same procedures as during the baseline survey ., Hence , duplicate Kato-Katz thick smears were prepared from each stool sample , examined under a microscope within 30–45 min by experienced laboratory technicians , and helminth eggs were counted and recorded for each species separately ., We adhered to the same quality control as during the baseline survey ., Children were randomly assigned to a single dose of albendazole ( 400 mg ) or mebendazole ( 500 mg ) , using a block randomization procedure ( six blocks each containing four treatment allocations ) , generated by an independent statistician who was not otherwise involved in the trial ., The sequence of blocks was determined using a random number table ., In addition , schools were decoded by a researcher to assign children either to albendazole or mebendazole ., Eligible children were randomly assigned and allocated to treatment by an epidemiologist ., Children and drug administrators were not blinded for drug treatment ., Laboratory personnel and clinicians monitoring the adverse events were blinded throughout the study ., Albendazole ( 400 mg; Albendazole® , South Korea ) was obtained from the Ministry of Health , Vientiane , Lao PDR ., Mebendazole ( 500 mg; Vermox® , Italy ) was donated by Johnson & Johnson Pharmaceuticals , provided through the Ministry of Health and the Ministry of Education , Vientiane , Lao PDR ., At treatment day , both groups received the drugs under direct medical supervision on an empty stomach ., Children were monitored for at least 3 hours after drug administration and asked to report for any drug-related adverse events using a standard questionnaire administered and graded by study physicians ., Data were double-entered and cross-checked in EpiData version 3 . 1 ( EpiData Association; Odense , Denmark ) ., Statistical analyses were performed with STATA , version 10 . 1 ( Stata Corp . ; College Station , TX , USA ) ., Efficacy was calculated for both intention-to-treat ( ITT ) and per-protocol ( PP ) analyses ., ITT analysis was based on the initial treatment intent ., PP analysis included only those children who had complete data records ( i . e . , quadruplicate Kato-Katz thick smear reading before and after treatment , and full treatment compliance ) ., Infections with hookworm , A . lumbricoides , T . trichiura , and O . viverrini were grouped into light , moderate , and heavy infections , according to WHO guidelines ( for STHs ) and cut-offs put forward by Maleewong and colleagues and WHO ( for O . viverrini ) 24 , 25 ., Infection intensity classifications are as follows: hookworm , 1–1 , 999 eggs per gram of stool ( EPG ) ( light ) , 2 , 000–3 , 999 EPG ( moderate ) , and ≥4 , 000 EPG ( heavy ) ; A . lumbricoides , 1–4 , 999 EPG ( light ) , 5 , 000–49 , 999 EPG ( moderate ) , and ≥50 , 000 EPG ( heavy ) ; and T . trichiura and O . viverrini , 1–999 EPG ( light ) , 1 , 000–9 , 999 EPG ( moderate ) , and ≥10 , 000 EPG ( heavy ) ., Primary outcome measures were CR and egg reduction rate ( ERR ) , the latter defined as the positive groups reduction of geometric mean ( GM ) fecal egg count at posttreatment , divided by the GM fecal egg count at pretreatment , multiplied by 100 ., Additionally , changes in class of infection intensities were determined following treatment ., Negative binomial regression was applied to compare ERRs observed between both treatment groups ., A Wilcoxen test was employed for the matched pairs analysis ., We determined egg reduction rate ratio ( ERRR ) and 95% confidence interval ( CI ) ., Pearsons χ2-test and Fishers exact test , as appropriate , were used to assess the baseline binary characteristics between the treatment arms ., Statistical significance was estimated using a likelihood ratio test ( LRT ) ., P-value below 5% was considered significant ., CONSORT checklist was followed to report on the trial ( see Checklist S1 ) ., Four hundred sixty-five school-aged children were enrolled in the baseline screening ., Two hundred children ( 43 . 0% ) , 130 boys and 70 girls with a parasitologically confirmed hookworm infection , were randomly assigned to one of the two treatments ., Data of these 200 children were included in the ITT analysis ., The remaining 265 children were excluded because they had no hookworm eggs in their stool ( n\u200a=\u200a235 ) or provided only a single stool sample ( n\u200a=\u200a30 ) ., Overall , 171 children ( 85 . 5% ) had complete baseline data , received treatment , and completed follow-up examinations , and hence PP analysis was performed on these children ., Twenty-nine children ( 14 . 5% ) were lost to follow-up , 18 in the mebendazole and 11 in the albendazole group ( Figure 1 ) ., The 171 children with complete data records were included in the primary analysis ., Their parents most commonly had completed primary school only ( 77 . 5% of parents for the albendazole group and 80 . 5% for the mebendazole group ) ., The most common profession of patients parents was farming with 49 . 4% and 62 . 2% for albendazole and mebendazole treatment groups , respectively ., The two groups were similar in terms of household assets , source of drinking water and consumption of cooked foods as well as raw fish ( data not shown ) ., More specifically , the consumption of raw fish was reported by 61 . 8% and 58 . 5% , respectively , and included dishes like “Pa Dek” ( fermented fish sauce ) , “Lap Pa” , and “Koy Pa” ( raw , fish-based dishes ) ., At baseline , characteristics of the two treatment groups were similar ( Table 1 ) , including age ( albendazole recipients: mean ( standard deviation , SD ) age 8 . 4 ( 2 . 1 ) years; mebendazole recipients: 8 . 7 ( 2 . 1 ) years ) , weight ( mean ( SD ) 23 . 8 ( 5 . 8 ) kg and 25 . 0 ( 5 . 9 ) kg , respectively ) , height ( mean ( SD ) 123 . 8 ( 11 . 0 ) cm and 126 . 9 ( 11 . 0 ) cm , respectively ) , and hemoglobin ( Hb ) concentration ( mean ( SD ) 11 . 8 ( 1 . 1 ) mg/dl and 11 . 9 ( 1 . 3 ) mg/dl , respectively ) ., In both treatment groups , most children were diagnosed with a light hookworm infection ( 82 . 0% ) , whereas the remaining children had moderate or heavy infection intensities ., The hookworm GM fecal egg counts in the mebendazole and albendazole groups were 707 . 0 and 859 . 1 EPG , respectively ( Table 2 ) ., The overall infection rates of A . lumbricoides , O . viverrini and T . trichiura were 34 . 0% , 48% and 45 . 0% , respectively ., O . viverrini GM fecal egg counts were 84 . 9 EPG ( albendazole ) and 120 . 8 EPG ( mebendazole ) ( Table 3 ) ., In the ITT analysis , the CRs of albendazole and mebendazole against hookworm infection were 32 . 0% and 15 . 0% , respectively ., Overall , 124 children ( 73% ) remained hookworm-egg positive; 68 receiving albendazole and 85 in the mebendazole treatment group ., Similar results were obtained with the PP analysis ( Table 2 ) ., A statistically significant difference was observed when comparing the observed CRs using albendazole vs . mebendazole ( OR\u200a=\u200a0 . 4; 95% CI 0 . 2–0 . 8; P\u200a=\u200a0 . 01 ) ., The hookworm GM fecal egg counts obtained at follow-up were 63 . 0 EPG in albendazole recipients and 147 . 3 EPG in mebendazole recipients ( ITT analysis 96 . 5 EPG and 210 EPG , respectively ) ., The respective ERRs for albendazole and mebendazole were 86 . 7% and 76 . 3% ( ERRR 1 . 0; 95%CI 0 . 7–1 . 6; P\u200a=\u200a0 . 90 ., In children with moderate infection intensities ( 2 , 000–3 , 999 EPG ) , the effect of albendazole and mebendazole was significantly different ( P\u200a=\u200a0 . 04 ) ., Table 3 shows the effect of albendazole and mebendazole against A . lumbricoides , T . trichiura , and O . viverrini ., At baseline , GM infection intensities of A . lumbricoides were 1 , 567 EPG in albendazole recipients and 1 , 584 EPG in mebendazole recipients ., Both albendazole and mebendazole treatments achieved high CRs above 90% and resulted in almost complete egg elimination ., The CRs of albendazole and mebendazole obtained against T . trichiura were 33 . 3% and 27 . 9% , respectively ., The respective ERRs were 67 . 0% and 66 . 0% ., No statistically significant difference was observed for CR and ERR between the two treatments ( OR\u200a=\u200a0 . 8; 95% CI 0 . 3–1 . 9; P\u200a=\u200a0 . 58 and ERRR\u200a=\u200a0 . 7; 95% CI 0 . 3–1 . 2 , P\u200a=\u200a0 . 22 ) ., Finally , CRs against O . viverrini achieved with albendazole and mebendazole were 33 . 3% and 24 . 2% , respectively ( OR\u200a=\u200a0 . 7; 95% CI 0 . 3–1 . 9; P\u200a=\u200a0 . 62 ) ., The respective ERRs were 82 . 1% and 78 . 2% ( ERRR\u200a=\u200a0 . 8; 95% CI 0 . 2–3 . 9 , P\u200a=\u200a0 . 78 ) ., Monitoring of children within 3 hours after drug administration revealed no drug-related adverse events , neither in the albendazole nor in the mebendazole group ., Hence , both treatments were well tolerated ., This current head-to-head comparison of single-dose albendazole vs . mebendazole against hookworm infection in Lao school-aged children – to our knowledge the first comparative trial in this Southeast Asian country – shows sobering results ., Indeed , the standard single oral doses of albendazole ( 400 mg ) and mebendazole ( 500 mg ) that are recommended for preventive chemotherapy targeting STHs 8 , 9 resulted in low CRs against hookworm infection ( 36 . 0% and 17 . 6% , respectively ) ., The respective ERRs were moderate , ( 86 . 7% and 76 . 3% ) ., A sizeable number of children were co-infected with A . lumbricoides , T . trichiura , and O . viverrini , which allowed us to determine the effect of albendazole and mebendazole against these helminth species ., With regard to A . lumbricoides , high efficacy of both drugs was confirmed against this helminth species 3 , 10 ., Our study also confirms the previously reported low efficacy of both drugs against T . trichiura 3 , 10 , 26 ., While the results obtained with mebendazole against hookworm and the efficacy observed with both drugs against A . lumbricoides and T . trichiura are in line with previous studies 20 , 27 , 28 and in agreement with overall CRs estimated through a meta-analysis 10 , the low CR ( 36 . 0% ) achieved with albendazole in the treatment of hookworm infection is somewhat surprising ., Indeed , in the aforementioned meta-analysis , randomized controlled trials of single-dose albendazole ( 400 mg ) revealed an overall CR against hookworm of 75% 10 ., The reasons for the considerably lower efficacy of albendazole observed in our study are unclear ., Quality control of drug samples performed in our laboratories revealed that disintegration , dissolution , and concentration of the albendazole tablets used in our trial were comparable to Zentel® ( data not shown ) ., The hookworm species ( and strains ) endemic in southern Lao PDR might be an explanation ., However , there is a paucity of information on which hookworm species is predominant in Southeast Asia ., Indeed , in our study setting the infection rates of the two hookworm species , A . duodenale and N . americanus , are not known ., Furthermore , recent studies documented that in Southeast Asia humans are at risk of acquiring Ancylostoma ceylanicum , which is endemic in dogs and cats of the region and its importance in humans might be underestimated 29 , 30 ., Hence , further analysis on the circulating parasite species is required to elucidate this issue ., In addition , day-to-day variability in hookworm egg counts from individuals is a well described phenomenon 31 ., Finally , the studys sample size is rather small and therefore a few incidental effects such as failure of some children to swallow the tablet correctly , might have contributed to low efficacy of albendazole for the treatment of hookworm infection ., To sum up , differences in strain and species susceptibilities , host factors , and co-infections with other helminths are factors that might all play a role in explaining treatment failures 28 , 32 ., Nevertheless , we cannot rule out that albendazole resistance is developing in our study setting ., To date , nematode resistance in humans has not been reported ., On the other hand , drug resistance is a major problem in veterinary public health 33 , 34 ., The development of broad spectrum anthelmintic resistance , in particular resistance of nematodes to benzimidazoles , has been recognized in ruminants for decades 34 , 35 ., Extensive studies on the underlying mechanisms of drug resistance have been carried out 36 ., Further investigations on failure of the drugs to completely cure the patients are necessary in our study setting to substantiate this suspicion ., It is interesting to note that the two drugs employed , even at single oral doses , showed some effect against O . viverrini ., Although CRs were low ( 24 . 2–33 . 3% ) , the moderate ERRs of 78 . 2–82 . 1% are encouraging ., At present , praziquantel is the drug of choice against opisthorchiasis 3 , 18 ., Studies carried out in the 1980s in O . viverrini-infected hamsters and patients infected with O . viverrini documented opisthorchicidal properties of albendazole and mebendazole 19 , 37 ., However , long treatment courses of up to 7 days were recommended in view of these initial laboratory and clinical findings ., Experiences with long treatment courses have been reported from a hospital-based randomized trial; albendazole given at dosages of 400 mg twice daily for 3 and 7 days resulted in CRs of 40% and 63% , respectively , and corresponding ERRs of 92% 19 ., Furthermore , mebendazole in dosages of 30 mg/kg daily for 3 or 4 weeks resulted in CRs of 94% against O . viverrini ., Long treatment courses compromise compliance , increase costs and are not feasible for large-scale community-based control , which might explain that albendazole and mebendazole were not further promoted for O . viverrini treatment 37 ., It should be noted that in our study Kato-Katz thick smears served as method for helminth diagnosis ., However , this diagnostic approach does not allow differentiating the eggs of O . viverrini from minute intestinal flukes 38 , 39 ., In addition , since the emphasis of our research was on hookworm , the efficacy of albendazole and mebendazole against other STHs and O . viverrini could not be compared with the appropriate sample sizes ., Finally , mostly light O . viverrini infections were present in our study and the sample of O . viverrini-infected patients was not representative of the overall community as hookworm infection was the leading selection criterion ., Hence , additional clinical investigations are warranted to assess the opisthorchicidal properties of albendazole and mebendazole in comparison to praziquantel ., Moreover , the anthelmintic drug tribendimidine 40 showed high CR and ERR against O . viverrini in a recent , open-label exploratory trial carried out in Lao PDR 41 ., It would therefore be interesting to conduct a four-arm study , comparing praziquantel ( treatment of choice ) with tribendimidine , albendazole , and mebendazole ., In conclusion , we have assessed the efficacy of standard single-dose regimens of albendazole and mebendazole against hookworm infection in school-aged children from Lao PDR and provide further evidence of the effects these two drugs have against other helminth species concurrently harbored in the human host ., Both drugs showed a similar profile , with low efficacy against hookworm and , additionally , low efficacy against T . trichiura , and high efficacy against A . lumbricoides ., The low efficacy of single-dose of albendazole against hookworm should be followed-up closely and further investigated as this drug is widely used for preventive chemotherapy against STHs and in combination with ivermectin in the current global effort to eliminate lymphatic filariasis ., The effects of the two drugs against O . viverrini warrant further investigations , in comparison with the current drug of choice praziquantel as well as tribendimidine . | Introduction, Methods, Results, Discussion | Albendazole and mebendazole are increasingly deployed for preventive chemotherapy targeting soil-transmitted helminth ( STH ) infections ., We assessed the efficacy of single oral doses of albendazole ( 400 mg ) and mebendazole ( 500 mg ) for the treatment of hookworm infection in school-aged children in Lao PDR ., Since Opisthorchis viverrini is co-endemic in our study setting , the effect of the two drugs could also be determined against this liver fluke ., We conducted a randomized , open-label , two-arm trial ., In total , 200 children infected with hookworm ( determined by quadruplicate Kato-Katz thick smears derived from two stool samples ) were randomly assigned to albendazole ( n\u200a=\u200a100 ) and mebendazole ( n\u200a=\u200a100 ) ., Cure rate ( CR; percentage of children who became egg-negative after treatment ) , and egg reduction rate ( ERR; reduction in the geometric mean fecal egg count at treatment follow-up compared to baseline ) at 21–23 days posttreatment were used as primary outcome measures ., Adverse events were monitored 3 hours post treatment ., Single-dose albendazole and mebendazole resulted in CRs of 36 . 0% and 17 . 6% ( odds ratio: 0 . 4; 95% confidence interval: 0 . 2–0 . 8; P\u200a=\u200a0 . 01 ) , and ERRs of 86 . 7% and 76 . 3% , respectively ., In children co-infected with O . viverrini , albendazole and mebendazole showed low CRs ( 33 . 3% and 24 . 2% , respectively ) and moderate ERRs ( 82 . 1% and 78 . 2% , respectively ) ., Both albendazole and mebendazole showed disappointing CRs against hookworm , but albendazole cured infection and reduced intensity of infection with a higher efficacy than mebendazole ., Single-dose administrations showed an effect against O . viverrini , and hence it will be interesting to monitor potential ancillary benefits of a preventive chemotherapy strategy that targets STHs in areas where opisthorchiasis is co-endemic ., Current Controlled Trials ISRCTN29126001 | Parasitic worms remain a public health problem in developing countries ., Regular deworming with the drugs albendazole and mebendazole is the current global control strategy ., We assessed the efficacies of a single tablet of albendazole ( 400 mg ) and mebendazole ( 500 mg ) against hookworm in children of southern Lao PDR ., From each child , two stool samples were examined for the presence and number of hookworm eggs ., Two hundred children were found to be infected ., They were randomly assigned to albendazole ( n\u200a=\u200a100 ) or mebendazole ( n\u200a=\u200a100 ) treatment ., Three weeks later , another two stool samples were analyzed for hookworm eggs ., Thirty-two children who were given albendazole had no hookworm eggs anymore in their stool , while only 15 children who received mebendazole were found egg-negative ., The total number of hookworm eggs was reduced by 85 . 3% in the albendazole and 74 . 5% in the mebendazole group ., About one third of the children who were co-infected with the Asian liver fluke Opisthorchis viverrini were cleared from this infection following albendazole treatment and about one forth in the mebendazole group ., Concluding , both albendazole and mebendazole showed disappointingly low cure rates against hookworm , with albendazole performing somewhat better ., The effect of these two drugs against O . viverrini should be studied in greater detail . | medicine, public health and epidemiology, parasitic diseases, food-borne trematodes, helminth infection, clinical epidemiology, hookworm infection, neglected tropical diseases, infectious disease control, ascariasis, infectious diseases, soil-transmitted helminths, parasitic intestinal diseases, epidemiology, hookworm, opisthorchiasis | null |
journal.pcbi.1003466 | 2,014 | Modelling Individual Differences in the Form of Pavlovian Conditioned Approach Responses: A Dual Learning Systems Approach with Factored Representations | Standard Reinforcement Learning ( RL ) 1 is a widely used normative framework for modelling conditioning experiments 2 , 3 ., Different RL systems , mainly Model-Based and Model-Free systems , have often been combined to better account for a variety of observations suggesting that multiple valuation processes coexist in the brain 4–6 ., Model-Based systems employ an explicit model of consequences of actions , making it possible to evaluate situations by forward inference ., Such systems best explain goal-directed behaviours and rapid adaptation to novel or changing environments 7–9 ., In contrast , Model-Free systems do not rely on internal models and directly associate values to actions or states by experience such that higher valued situations are favoured ., Such systems best explain habits and persistent behaviours 9–11 ., Of significant interest , learning in Model-Free systems relies on a computed reinforcement signal , the reward prediction error ( RPE ) ., This signal parallels the observed shift of dopamine neurons response from the time of an initially unexpected reward – an outcome that is better or worse than expected – to the time of the conditioned stimulus that precedes it , which , in Pavlovian conditioning experiments , is fully predictive of the reward 12 , 13 ., However recent work by Flagel et al . 14 , raises questions about the exclusive use of classical RL Model-Free methods to account for data in Pavlovian conditioning experiments ., Using an autoshaping procedure , a lever-CS was presented for 8 seconds , followed immediately by delivery of a food pellet into an adjacent food magazine ., With training , some rats ( sign-trackers; STs ) learned to rapidly approach and engage the lever-CS ., However , others ( goal-trackers; GTs ) learned to approach the food magazine upon CS presentation , and made anticipatory head entries into it ., Furthermore , in STs , phasic dopamine release in the nucleus accumbens , measured with fast scan cyclic voltammetry , matched RPE signalling , and dopamine was necessary for the acquisition of a sign-tracking CR ., In contrast , despite the fact that GTs acquired a Pavlovian conditioned approach response , this was not accompanied with the expected RPE-like dopamine signal , nor was the acquisition of a goal-tracking CR blocked by administration of a dopamine antagonist ( see also 15 ) ., Classical dual systems models 16–19 should be able to account for these behavioural and pharmacological data , but the physiological data are not consistent with the classical view of RPE-like dopamine bursts ., Based on the observation that STs and GTs focus on different stimuli in the environment , we suggest that the differences observed in dopamine recordings may be due to an independent valuation of each stimulus ., In classical RL , valuation is usually done at the state level ., Stimuli , embedded into states – snapshots of specific configurations in time – , are therefore hidden to systems ., In this case , it would prevent dealing separately with the lever and the magazine at the same time ., However , such data may still be explained by a dual systems theory , when extended to support and benefit from factored representations; that is , learning the specific value of stimuli independently from the states in which they are presented ., In this paper , we present and test a model using a large set of behavioural , physiological and pharmacological data obtained from studies on individual variation in Pavlovian conditioned approach behaviour 14 , 20–25 ., It combines Model-Free and Model-Based systems that provide the specific components of the observed behaviours 26 ., It explains why inactivating dopamine in the core of the nucleus accumbens or in the entire brain results in blocking specific components and not others 14 , 25 ., By weighting the contribution of each system , it also accounts for the full spectrum of observed behaviours ranging from one extreme – sign-tracking – to the other 26 – goal-tracking ., Above all , by extending classical Model-Free methods with factored representations , it potentially explains why the lever-CS and the food magazine might acquire different motivational values in different individuals , even when they are trained in the same task 22 ., It may also account for why the RPE-like dopaminergic responses are observed in STs but not GTs , and also the differential dependence on dopamine 14 ., Not only have Flagel et al . 14 provided behavioural data but they also provide physiological and pharmacological data ., This raises the opportunity to challenge the model at different levels , as developed in the current and next sections ., Using Fast Scan Cyclic Voltammetry ( FSCV ) in the core of the nucleus accumbens they recorded the mean of phasic dopamine ( DA ) signals upon CS ( lever ) and US ( food ) presentation ., It was observed that depending on the subgroup of rats , distinct dopamine release patterns emerge ( see Figure 7 A , B ) during Pavlovian training ., STs display the classical propagation of a phasic dopamine burst from the US to the CS over days of training and the acquisition of conditioned responding ( see Figure 7 A ) ., This pattern of dopamine activity is similar to that seen in the firing of presumed dopamine cells in monkeys reported by Schultz and colleagues 12 and interpreted as an RPE corresponding to the reinforcement signal of Model-Free RL systems 1 ., In GTs , however , a different pattern was observed ., Initially there were small responses to both the CS and US , of which the amplitudes seemed to follow a similar trend over training ( see Figure 7 B ) ., By recording the mean of the RPEs computed in the Feature-Model-Free system during the autoshaping simulation ( i . e . only fitted to behavioural data ) , the model can still qualitatively reproduce the different patterns observed in dopamine recordings for STs and GTs ( see Figure 7 C , D ) ., For STs , the model reproduces the progressive propagation of from the US to the CS ( see Figure 7 C ) ., For GTs , it reproduces the absence of such propagation ., The RPE at the time of the US remains over training , while a also appears at the time of the CS ( see Figure 7 D ) ., In the model , such discrepancy is explained by the difference in the values that STs and GTs use for the computation of RPEs at the time of the CS and the US ., STs , by repeatedly focusing on the lever , propagate the total value of food to the lever and end up having a unique at the unexpected lever appearance only ., By contrast , by repeatedly focusing on the magazine during the lever appearance but , as all rats , also from time to time during ITI , GTs revise the magazine value multiple times , positively just after food delivery and negatively during ITI ., Such revisions lead to a permanent discrepancy between the expected and observed value , i . e . a permanent , at lever appearance and food delivery , when engaging with the magazine ., The key mechanism to reproduce these results resides in the generalization capacities of the Feature-Model-Free system ., Based on features rather than states , feature-values are to be used , and therefore revised , at different times and states of the experiment , favouring the appearance of RPEs ., Variants 2 , 3 and 4 relying on classical Model-Free systems are unable to reproduce such results ( see Figure S3 ) ., By using values over abstract states rather than stimuli , it makes it impossible to only revise the value of the magazine during ITI ., Therefore , given the deterministic nature of the MDP , we observe a classical propagation of RPEs in all pathways up to the appearance of the lever ., Combining Model-Based and Model-Free systems has previously been successful in explaining the shift from goal-directed to habitual behaviours observed in instrumental conditioning 17–19 , 33 , 34 ., However , few models based on the same concept have been developed to account for Pavlovian conditioning 16 ., While the need for two systems is relevant in instrumental conditioning given the distinct temporal engagement of each system , such a distinction has not been applied to Pavlovian phenomena ( but see recent studies on orbitofrontal cortex 37–39 ) ., The variability of behaviours and the need for multiple systems have been masked by focusing on whole populations and , for the most part , ignoring individual differences in studies of Pavlovian conditioning ., The nature of the CS is especially important , as many studies of Pavlovian conditioned approach behaviour have used an auditory stimulus as the CS , and in such cases only a goal-tracking CR emerges in rats 40 , 41 ., As expected from the behavioural data , combining two learning systems was successful in reproducing sign- and goal-tracking behaviours ., The Model-Based system , learning the structure of the task , favours systematic approach towards the food magazine , and waiting for food to be delivered , and hence the development of a goal-tracking CR ., The Feature-Model-Free system , directly evaluating features by trials and errors , favours systematic approach towards the lever , a full predictor of food delivery , and hence the development of a sign-tracking CR ., Moreover , utilizing the Feature-Model-Free system to represent sign-tracking behaviour yields results consistent with the pharmacological data ., Disrupting RPEs , which reflects the effects of flupentixol on dopamine , blocks the acquisition of a sign-tracking CR , but not a goal-tracking CR ., The model does not make a distinction between simple approach behaviour versus consumption-like engagement , as reported for both STs and GTs 23 , 24 ., However given that such engagement results from the development of incentive salience 23 , 24 , the values learned by the Feature-Model-Free system to bias behaviour towards stimuli attributed with motivational value are well-suited to explain such observations ., The higher motivational value attributed to the lever by STs relative to GTs can also explain why the lever-CS is a more effective conditioned reinforcer for STs than for GTs 22 ., Importantly , none of the systems are dedicated to a specific behaviour , nor rely on a priori information to guide their processes ., The underlying mechanisms increasingly make one behaviour more pronounced than the other through learning ., Each system contributes to a certain extent to sign- and goal-tracking behaviour ., This property is emphasized by the weighted sum integration of the values computed by each system before applying the softmax action-selection mechanism ., The variability of behaviours in the population can then be accounted for by adjusting the weighting parameter from ( i . e . favouring sign-tracking ) to ( i . e . favouring goal-tracking ) ., This suggests that the rats actions result from some combination of rational and impulsive processes , with individual variation contributing to the weight of each component ., The integration mechanism is directly inspired by the work of Dayan et al . 16 and as the authors suggest , the parameter may fluctuate over time , making the contribution of the two systems vary with experience ., In contrast to their model , however , the model presented here does not assign different goals to each system ., Thus , the current model is more similar to their previous model 17 , which uses another method for integration ., A common alternative to integration when using multiple systems 17 , 18 , 35 is to select at each step , based on a given criterion ( certainty , speed/accuracy trade-off , energy cost ) , a single system to pick the next action ., Such switch mechanism does not fit well with the present model , given that it would be interpreted as if actions relied sometimes only on motivational values ( i . e . Feature-Model-Free system ) and sometimes only on a rational analysis of the situation ( i . e . Model-Based system ) ., It also does not fit well with pharmacological observation that STs do not express goal-tracking tendencies in the drug-free test session following systemic-injections of flupentixol 14 , as Flagel et al . stated , “sign-tracking rats treated with flupentixol did not develop a goal-tracking CR” ., Classical RL algorithms used in neuroscience 16–18 , 35 , designed mainly to account for instrumental conditioning , work at the state level ., Tasks are defined as graphs of states , and corresponding models are unaware of any similarity within states ., Therefore , any subsequent valuation process cannot use any underlying structure to generalize updates to states that share stimuli ., Revising the valuation process to handle features rather than states per se , makes it possible to attribute motivational values to stimuli independently of the states in which they are presented ., Recent models dedicated to Pavlovian conditioning 36 , 42–46 usually represent and process stimuli independently and can be said to use factored representations , a useful property to account for phenomena such as blocking 47 or overexpectation 48 ., In contrast to the present model , while taking inspiration from RL theory ( e . g . using incremental updates ) , these models are usually far from the classical RL framework ., Of significant difference with the present study , most of these models tend to describe the varying intensity of a unique conditioned response and do not account for variations in the actual form of the response , as we do here ., In such models , the magazine would not be taken into account and/or taken as part of the context , making it unable to acquire a value for itself nor be the focus of a particular response ., In RL theory , factorization is mainly evoked when trying to overcome the curse of dimensionality 49 ( i . e . standard algorithms do not scale well to high dimensional spaces and require too much physical space or computation time ) ., Amongst methods that intend to overcome this problem are value function approximations and Factored Reinforcement Learning ., Value function approximations 35 , 50 , 51 attempt to split problems into orthogonal subproblems making computations easier and providing valuations that can then be aggregated to estimate the value of states ., Factored Reinforcement Learning 52–54 attempts to find similarities between states so that they can share values , reducing the physical space needed and relies on factored Markov Decision Processes ., We also use factored Markov Decision processes , hence the “factored” terminology ., However , our use of factored representations serves a different purpose ., We do not intend to build a compact value-function nor infer the value of states from values of features but rather make these values compete in the choice for the next action ., Taking advantage of factored representations into classical RL algorithms is at the very heart of the present results ., By individually processing stimuli within states ( i . e . in the same context , at the same time and same location ) and making them compete , the Feature-Model-Free system favours a different policy – oriented towards engaging with the most valued stimuli – ( sign-tracking ) than would have been favoured by classical algorithms such as Model-Based or Model-Free systems ( goal-tracking ) ., Hence , combining a classical RL algorithm with the Feature-Model-Free system enables the model to reproduce the difference in behaviours observed between STs and GTs during an autoshaping procedure ., Moreover , by biasing expected optimal behaviours towards cues with motivational values ( incentive salience ) , it is well suited to explain the observed commitment to unnecessary and possibly counter-productive actions ( see also 16 , 55 , 56 ) ., Most of all , it enables the model to replicate the different patterns of dopamine activity recorded with FSCV in the core of the nucleus accumbens of STs and GTs ., The independent processing of stimuli leads to patterns of RPE that match those of dopamine activity for STs – a shift of bursts from the US to the CS; and in GTs – a persistence of bursts at both the time of the US and the CS ., By combining the two concepts of dual learning systems and factored representations in a single model , we are able to reproduce individual variation in behavioural , physiological and pharmacological effects in rats trained using an autoshaping procedure ., Interestingly , our approach does not require a deep revision of mechanisms that are extensively used in our current field of research ., While Pavlovian and instrumental conditioning seem entangled in the brain 57 , the two major concepts on which rely their respective models , dual learning systems and factored representations , have to our knowledge never been combined into a single model in this field of research ., This approach could contribute to the understanding of interactions between these two classes of learning , such as CRE or Pavlovian-Instrumental Transfer ( PIT ) , where motivation for stimuli acquired via Pavlovian learning modulates the expression of instrumental responses ., Interestingly , the Feature-Model-Free system nicely fits with what would be expected from a mechanism contributing to general PIT 58 ., It is focused on values over stimuli without regard to their nature 58 , it biases and interferes with some more instrumental processes 55 , 56 , 58 and it is hypothesized to be located in the core of the nucleus accumbens 58 ., It would thus be interesting to study whether future simulations of the model could explain and help better formalize these aspects of PIT ., We do not necessarily imply that instrumental and Pavlovian conditioning might rely on a unique model ., Rather , we propose that if they were the results of separated systems , they should somehow rely on similar representations and valuation mechanisms , given the strength of the observed interactions ., The proposed model explains the persistent dopamine response to the US in GTs over days of training as a permanent RPE due to the revision of the magazine value during each ITI ., Therefore , a prediction of the model is that shortening the ITI should reduce the amplitude of this burst ( i . e . there should be less time to revise the value and reduce the size of the RPE ) ; whereas increasing the ITI should increase the amplitude of this burst ., Removing the food dispenser during ITI , similar to theoretically suppressing the ITI , should make this same burst disappear ., Studying physiological data by grouping them given the duration of the preceding ITI might be sufficient , relatively to noise , to confirm that its duration impacts the amplitude of dopamine bursts ., In the current experimental procedure , the ITI is indeed randomly picked in a list of values with an average of 90 sec ., Moreover , reducing ITI duration should lead to an increase of the tendency to goal-track in the overall population ., Indeed , with a higher value of the food magazine , the Feature-Model-Free system would be less likely to favour sign-tracking over goal-tracking CR ., The resulting decrease in sign-tracking in the overall population would be consistent with findings of previous works 59–62 , where a shorter ITI reduces the observed performance in the acquisition of sign-tracking CRs ., Alternatively , it would also be interesting to examine the amplitude of dopamine bursts during the ITI ( especially when exploring the food magazine ) , to determine whether or not physiological responses during this period affect the outcome of the conditioned response ., It would be interesting to split physiological data not only between STs and GTs but also between the stimuli on which the rats started and/or ended focusing on during CS presentation at each trial ., This would help to confirm that the pattern of dopamine activity is indeed due to a separate valuation of each stimuli ., We would predict that at the time of the US , dopamine bursts during engagement with the lever should be small relatively to dopamine bursts during engagement with the magazine ., Moreover , comparing dopamine activity at the time of the CS when engaging with the lever versus the magazine could help elucidate which update mechanism is being used ., If activity differs , this would suggest that the model should be revised to use SARSA-like updates , i . e . taking into account the next action in RPE computation ., Such a question has already been the focus of some studies on dopamine activity 63–65 ., There is no available experimental data for the phasic dopaminergic activity of the intermediate group ., The model predicts that such a group would have a permanent phasic dopamine burst , i . e . RPE , at US and a progressively appearing burst at CS ( see Figure S6 ) ., Over training , the amplitude of the phasic dopamine burst at US should decrease until a point of convergence , while at the mean time the response at CS should increase until reaching a level higher than the one observed at US ., However , one must note , that the fitting of the intermediate group is not as good as for STs or GTs , as it regroups behaviours that range from sign-tracking to goal-tracking , such that this is a weak prediction ., There is the possibility that regularly presenting the magazine or the lever could , without pairing with food , lead to responses that are indistinguishable from CRs ., However , ample evidence suggests that the development of a sign-tracking or goal-tracking CR is not due to this pseudoconditioning phenomenon , but rather a result of learned CS-US associations ., That is , experience with lever-CS presentations or with food US does not account for the acquisition of lever-CS induced directed responding 22 , 66 ., Nonetheless , it should be noted that the current model cannot distinguish between pseudoconditioning CR-like responses and sign-tracking or goal-tracking behaviours ., This would require us to introduce more complex MDPs that embed the ITI and can more clearly distinguish between approach and engagement ., The Feature-Model-Free system presented in this article was designed as a proof of concept for the use of factored representations in computational neuroscience ., In its present form it updates the value of one feature ( the focused one ) at a time , and this is sufficient to account for much of the experimental data ., It does not address whether multiple features could be processed in parallel , such that multiple synchronized , but independently computed , signals would update distinct values relative to the attention paid to the associated features ., Further experiments should be performed to confirm this hypothesis ., Subsequently , using factored representations in the Model-Based system was not necessary to account for the experimental data and the question remains whether explaining some phenomena would require it ., While using factored representations , our approach still relies on the discrete-time state paradigm of classical RL , where updates are made at regular intervals ., Although such simplification can explain the set of data considered here , one would need to extend this to continuous time if one would like to also model experimental data where rats take more or less time to initiate actions that can vary in duration 14 ., The present model , which does not take timing into consideration , cannot account for the fact that STs and GTs both come to approach their preferred stimuli faster and faster as a function of training nor does it make use of the variations of ITI duration ., Our attempt to overcome this limitation using the MDP framework was unsuccessful ., Focusing on features , it becomes more tempting to deal with the timing of their presence , a property that is known to be learned and to have some impact on behaviours 61 , 67–69 ., Moreover , in the current model , we did not attempt to account for the conditioned orienting responses ( i . e . orientation towards the CS ) that both STs and GTs exhibit upon CS presentation 25 ., However , we hypothesize that such learned orienting responses could be due to state discrimination mechanisms that are not included in the model , and would be better explained with partial observability and actions dedicated to collect information ., This is beyond the scope of the current article , but is of interest for future studies ., As evident by the only partial reproduction of the flupentixol effects on the expression of sign- and goal-tracking behaviours , the model is limited by the use of the softmax action-selection mechanism , which is widely used in computational neuroscience 16–19 , 32 , 34–36 ., In the model , all actions are equal – there is no action with a specific treatment – and the action-selection mechanism necessarily selects an action at each time step ., Any reduction in the value of one action favours the selection of all other actions in proportion to their current associated values ., In reality , however , blocking the expression of an action would certainly lead mainly to inactivity rather than necessarily picking the alternative and almost never expressed action ., One way of improving the model in this direction could be to replace the classical softmax function by a more realistic model of action selection in the basal ganglia ( e . g . 70 ) ., In such a model , no action is performed when no output activity gets above a certain threshold ., Humphries et al . 32 have shown that changing the exploration level in a softmax function can be equivalent to changing the level of tonic dopamine in the basal ganglia model of Gurney et al . 70 ., Interestingly , in the latter model , reducing the level of tonic dopamine results in difficulty in initiating actions and thus produces lower motor behaviour , as is seen in Parkinsonian patients and as can be seen in rats treated with higher doses of flupentixol 14 ., Thus a natural sequel to the current model would be to combine it with a more realistic basal ganglia model for action selection ., We simulated the effect of flupentixol as a reduction of the RPE in the learning processes of Model-Free systems to parallel its blockade of the dopamine receptors ., While this is sufficient to account for the pharmacological results previously reported 14 , it fails to account for some specific aspects that have more recently emerged ., Mainly , it is unable to reproduce the instant decreased engagement observed at the very first trial after post-training local injections of flupentixol 25 ., Our current approach requires re-learning to see any impact of flupentixol ., A better understanding of the mechanisms that enable instant shifts in motivational values , by shifts in the motivational state 71 or the use of drugs 14 , 25 , might be useful to extend the model on such aspects ., We also tried to model the effect of flupentixol on RPEs with a multiplicative effect , as it would have accounted for an instant impact on behaviour ., However , it failed to account for the effects of flupentixol on learning of the sign-tracking CRs , as a multiplicative effect only slowed down learning but did not disrupt it ., How to model the impact of flupentixol , and dopamine antagonists or drugs such as cocaine remains an open question ( e . g . see 72 , 73 ) ., Finally , our work does not currently address the anatomical counterpart of at the heart of the model , nor the regions of the brain that would match the current Model-Based system and the Feature-Model-Free system ., Numerous studies have already discussed the potential substrates of Model-Based/Model-Free systems in the prefrontal cortex/dorsolateral striatum 74 , or the dorsomedial and dorsolateral striatum 33 , 75–78 ., The weighted sum integration may suggest a crossed projection of brains regions favouring sign- and goal-tracking behaviours ( Model-Based and Feature-Model-Free systems ) into a third one ., We postulate there is a difference in strength of “connectivity” between such regions in STs vs GTs 79 ., Further , one might hypothesize that the core of the nucleus accumbens contributes to the Feature-Model-Free system ., The integration and action selection mechanisms would naturally fit within the basal ganglia , stated to contribute to such functions 32 , 80–82 ., Here we have presented a model that accounts for variations in the form of Pavlovian conditioned approach behaviour seen during autoshaping in rats; that is , the development of a sign-tracking vs goal-tracking CR ., This works adds to an emerging set of studies suggesting the presence and collaboration of multiple RL systems in the brain ., It questions the classical paradigm of state representation and suggests that further investigation of factored representations in RL models of Pavlovian and instrumental conditioning experiments may be useful ., In the classical reinforcement learning theory 1 , tasks are usually described as Markov Decision Processes ( MDPs ) ., As the proposed model is based on RL algorithms , we use the MDP formalism to computationally describe the Pavlovian autoshaping procedure used in all simulations ., An MDP describes the interactions of an agent with its environment and the rewards it might receive ., An agent being in a state can execute an action which results in a new state and the possible retrieval of some reward ., More precisely , an agent can be in a finite set of states , in which it can perform a finite set of discrete actions , the consequences of which are defined by a transition function , where is the probability distribution of reaching state doing action in state ., Additionally , the reward function is the reward for doing action in state ., Importantly , MDPs should theoretically comply with the Markov property: the probability of reaching state should only depend on the last state and the last action ., An MDP is defined as episodic if it includes at least one state which terminates the current episode ., Figure 1 shows the deterministic MDP used to simulate the autoshaping procedure ., Given the variable time schedule ( 30–150s ) and the net difference observed in behaviours in inter-trial intervals , we can reasonably assume that each experimental trial can be simulated with a finite horizon episode ., The agent starts from an empty state ( ) where there is nothing to do but explore ., At some point the lever appears ( ) and the agent must make a critical choice: It can either go to the lever ( ) and engage with it ( ) , go to the magazine ( ) and engage with it ( ) or just keep exploring ( , ) ., At some point , the lever is retracted and food is delivered ., If the agent is far from the magazine ( , ) , it first needs to get closer ., Once close ( ) , it consumes the food ., It ends in an empty state ( ) which symbolizes the start of the inter-trial interval ( ITI ) : no food , no lever and an empty but still present magazine ., The MDP in Figure 1 is common to all of the simulations and independent of the reinforcement learning systems we use ., STs should favour the red path , while GTs should favour the shorter blue path ., All of the results rely mainly on the action taken at the lever appearance ( ) , when choosing to go to either the lever , the magazine , or to explore ., Exploring can be understood as not going to the lever nor to the magazine ., To fit with the requirements of the MDP framework , we introduce two limitations in our description , which also simplify our analyses ., We assume that engagement is necessarily exclusive to one or no stimulus , and we make no use of the precise timing of the procedure – the ITI duration nor the CS duration – in our simulations ., The model relies on the architecture shown in Figure 2 ., The main idea is to combine the computations of two distinct reinforcement learning systems to define what behavioural response is chosen at each step ., Given the modular architecture of the model , we were able to test different combinations of RL systems ., Their analysis underlined | Introduction, Results, Discussion, Methods | Reinforcement Learning has greatly influenced models of conditioning , providing powerful explanations of acquired behaviour and underlying physiological observations ., However , in recent autoshaping experiments in rats , variation in the form of Pavlovian conditioned responses ( CRs ) and associated dopamine activity , have questioned the classical hypothesis that phasic dopamine activity corresponds to a reward prediction error-like signal arising from a classical Model-Free system , necessary for Pavlovian conditioning ., Over the course of Pavlovian conditioning using food as the unconditioned stimulus ( US ) , some rats ( sign-trackers ) come to approach and engage the conditioned stimulus ( CS ) itself – a lever – more and more avidly , whereas other rats ( goal-trackers ) learn to approach the location of food delivery upon CS presentation ., Importantly , although both sign-trackers and goal-trackers learn the CS-US association equally well , only in sign-trackers does phasic dopamine activity show classical reward prediction error-like bursts ., Furthermore , neither the acquisition nor the expression of a goal-tracking CR is dopamine-dependent ., Here we present a computational model that can account for such individual variations ., We show that a combination of a Model-Based system and a revised Model-Free system can account for the development of distinct CRs in rats ., Moreover , we show that revising a classical Model-Free system to individually process stimuli by using factored representations can explain why classical dopaminergic patterns may be observed for some rats and not for others depending on the CR they develop ., In addition , the model can account for other behavioural and pharmacological results obtained using the same , or similar , autoshaping procedures ., Finally , the model makes it possible to draw a set of experimental predictions that may be verified in a modified experimental protocol ., We suggest that further investigation of factored representations in computational neuroscience studies may be useful . | Acquisition of responses towards full predictors of rewards , namely Pavlovian conditioning , has long been explained using the reinforcement learning theory ., This theory formalizes learning processes that , by attributing values to situations and actions , makes it possible to direct behaviours towards rewarding objectives ., Interestingly , the implied mechanisms rely on a reinforcement signal that parallels the activity of dopamine neurons in such experiments ., However , recent studies challenged the classical view of explaining Pavlovian conditioning with a single process ., When presented with a lever whose retraction preceded the delivery of food , some rats started to chew and bite the food magazine whereas others chew and bite the lever , even if no interactions were necessary to get the food ., These differences were also visible in brain activity and when tested with drugs , suggesting the coexistence of multiple systems ., We present a computational model that extends the classical theory to account for these data ., Interestingly , we can draw predictions from this model that may be experimentally verified ., Inspired by mechanisms used to model instrumental behaviours , where actions are required to get rewards , and advanced Pavlovian behaviours ( such as overexpectation , negative patterning ) , it offers an entry point to start modelling the strong interactions observed between them . | behavioral neuroscience, computational neuroscience, biology, neuroscience, learning and memory | null |
journal.pcbi.1004698 | 2,016 | Optimal Information Representation and Criticality in an Adaptive Sensory Recurrent Neuronal Network | The anatomical abundance of lateral interactions 1 , 2 between neurons of the local cerebral circuit ( referred in this text as recurrent connections ) suggest they play a fundamental role in cortical function ., Indirect physiological evidence of their involvement in memory 3 , 4 , sensory processing 5 and in other brain functions 6 , 7 reinforces this notion ., Various models have been put forward in an attempt to explain the role of these lateral connections , however , an agreed framework is still missing and the topic is still far from being concluded ., In the narrower scope of early visual cortex , some studies have related the role of recurrent connectivity to orientation tuning and contrast invariance 8–10 ., Others have suggested a role in generating the accurate firing rates common to spontaneous activity 11 ., An additional aspect of recurrently connected networks ( relative to networks connected by feedforward links only ) involves their dynamic properties ., Networks with recurrent connections have been shown to form associative-memory related attractor states12 , 13 , exhibit self-organization leading to “neuronal avalanches” 14 , 15 , and in general , have the potential to exhibit critical dynamics 16–18 ., The idea that brain areas may operate near criticality was proposed on theoretical grounds by several authors in the past 18–22 ., There is also a growing bulk of recent experimental evidence supporting it 14 , 15 , 23–26 ( for reviews on near criticality in the brain see 16 , 27 ) ., Beggs and Plenz 14 , 15 demonstrated that neural activity in acute slices and in slice cultures is organized in neural avalanches , whose size obeys a power law distribution ., They interpreted their results in terms of critical branching processes 28 ., Further work 23 showed that neuronal avalanches also appear in the spontaneous cortical activity of awake monkeys and in large scale human brain activity ( e . g . 29 , 30 ) ., It was also demonstrated in slice cultures that the dynamical range of the network is maximized near the critical point 24 ., Although these dynamic properties have by now been well established , only few papers in the neuroscience literature have so far attempted to link them to concrete brain functions , such as the function of the visual system ., A central question regarding recurrent interactions , which has not yet been properly addressed , is how they evolve to facilitate the network’s computational capacity and what principles govern this evolution ., Their optimal pattern within the network also remains unknown ., In this work , we address these issues using a first-principle information theoretic approach , namely using the principle of maximum information preservation ( also known as ‘infomax’ 31 ) ., This principle has been successfully implemented in a variety of computational neuroscience studies ., Bell & Sejnowski 32 extended it to nonlinear output neurons implementing ICA ( Independent Component Analysis ) to achieve blind source separation ., Later , they showed that the independent components of natural scenes are Gabor-like edge filters 33 ., Tanaka et al 34 have demonstrated that the characteristics of orientation selectivity in V1 can be acquired by self-organization of recurrent neural networks according to Infomax learning ., This work was recently extended by Hayakawa et al 35 to reveal a biologically plausible infomax learning algorithm ., The present work can be seen as a further extension of these earlier efforts , studying how the gradual development of a network’s recurrent interactions may optimize the representation of input stimuli ., Unsupervised learning is applied in training networks to maximize mutual information between the input layer and an overcomplete recurrently connected output layer ., The evolving pattern of recurrent interactions is investigated in a model of a hypercolumn in primary visual cortex , considered the base functional unit of V1 , receiving input from both eyes , in a full representation of all possible orientations ., Various constellations of input stimuli and network connectivity are examined , in aim of studying their relationship with different network measures ., Methods to evaluate the optimal pattern of recurrent interactions in a neural network model and its dependence on the statistics of the external inputs were extended from Shriki et al . 36 ., We first provide an analytical and numerically simulated account of a toy hypercolumn network model ., Subsequently , a more ecological network is studied , in which natural scenes are used as input for training the network ., These models allow us to compare the emerging network’s properties with those arising from earlier empirical and theoretical work ., The basic network model consists of two layers of neurons , N neurons at the input layer and M neurons at the output layer ( Fig 1A ) , where M ≥ N . Thus , the network deterministically maps a low dimensional input space into a manifold in a higher-dimensional output space ., Such a representation , which contains more output components than input components , is termed overcomplete ., The feedforward interactions are described by the M × N matrix W and the recurrent interactions by the M × M matrix K . During the presentation of each input sample , the input components xi are fixed ., The dynamics of the output neurons are given by, τdsidt=−si+g ( Σj=1NWijxj+Σk=1MKikSk ) , i=1 , … , M, ( 1 ), Where g is some nonlinear squashing function and τ is a characteristic time scale ( here we set τ = 1 and g was taken to be the logistic function , g ( x ) = 1/ ( 1+e−x ) ) ., We assume that the activities of the output neurons reach equilibrium after some time and define the output as the steady-state pattern of activity ., For the cases we studied , numerical simulations of the network dynamics indeed stabilized and proved this assumption to be consistent ., The steady-state responses are given by, si=g ( Σj=1NWijxj+Σk=1MKikSk ) , i=1 , … , M, ( 2 ) To evaluate the neuronal representation of the external inputs we used the mutual information between the input and output of the network 37 ., More specifically , the mutual information between the input vector , x , and the output vector , s , can be expressed as the difference between the entropy of the output and the conditional entropy of the output given the input ., The conditional entropy can also be viewed as the entropy of the output noise ., Here , the network response is a deterministic function of the input , and thus the mutual information depends only on the entropy of the outputs ., As shown in 36 , maximizing the output entropy ( and therefore the mutual information ) is equivalent to minimizing the following objective function:, ε=−12〈ln\u2009det ( χTχ ) 〉x=−12Tr〈ln ( χTχ ) 〉x, ( 3 ), where xij=∂si∂xj is the Jacobian matrix of the transformation and reflects the sensitivity of the output units to changes in the input units ., We also refer to this matrix as the susceptibility matrix as it is analogous to the susceptibility of physical systems to external fields ., The adaptive parameters of the algorithm are the sets of feedforward and recurrent interactions , Wij and Kij ., The learning rules for these parameters are derived from this objective function using the gradient decent method , as shown in 36 ., Here we focus only on the recurrent interactions ., The gradient descent learning rule for the recurrent interactions is:, ΔK=−η∂ε∂K=η〈 ( χΓ ) T+ϕTasT〉, ( 4 ), Where η is the learning rate , the matrix ϕ is given by ϕ = ( G−1−K ) −1 and satisfies χ = ϕW , the matrix G is defined as gk\xa0→\xa0gi , the matrix Γ is defined as Γ = ( χTχ ) −1χTϕ and the components of the vector a are given by ak=χΓkkgk″ ( gk′ ) 3 ., The triangular brackets denote averaging over the input samples ., We defined several measures to characterize the behavior of the network and gain further insight into its dynamics ., As described in the Results section , after the learning process converges , the networks tend to operate near a critical point ., Thus , it is helpful to define metrics that may behave differently when the networks approach that critical point ., One such measure is the time it takes the recurrent network dynamics to reach steady-state—the convergence time ., Many dynamical systems exhibit a slow-down of the dynamics near critical points , often termed critical slowing down 38 ., Thus , a substantial increase in the convergence time may indicate that the system is close to a critical point ., To gain insight in the present context , we note that near a steady state , the linearized dynamics ( in vector notation ) are given by τd ( δs ) dt=−I−GKδs ., The inverse of the matrix I−GK appears also in the expression for the Jacobian matrix , which determines the objective function ., Optimizing the objective function leads to very large eigenvalues in the Jacobian matrix ( high-susceptibility ) , and therefore , the eigenvalues that dominate the dynamics become very small , which manifests as slowing down ., To estimate the convergence time , we defined a criterion for stability of the neuronal activities and measured the time it takes the network to satisfy this criterion ., This stability criterion means that for each neuron in the network , the difference in its activity between the current time step and the previous time step is smaller than a predefined small number ., When the network becomes supercritical , it converges onto attractor states , which reflect the underlying connectivity ., In the context of orientation tuning , which we study here , a natural measure to quantify this behavior is the population vector 39 ., Each neuron is associated with a complex number ., The magnitude of the number is the activity of this neuron and the phase is set according to the preferred angle or orientation of the neuron ( in the case of preferred orientation , the orientation is multiplied by 2 , to span the range from 0° to 360° ) ., Given a pattern of activity in the network , these complex numbers are summed to yield a resultant complex number , termed the population vector ., When the network response is uniform , the magnitude of the population vector is 0 ., When the network response peaks at some orientation , the magnitude of the population vector is finite ., Similar to previous papers concerning training of networks over natural scenes 33 , we used photos involving forest scenes or single trees and leaves ., The photos were converted to grayscale byte value of 0 to 255 and then”cut” into patches of 25-by-25 pixels ., Each patch was represented as a vector with 625 components ., Using PCA ( Principal Component Analysis ) , the dimensionality of the images was reduced from 625 to 100 ., The inputs were also whitened by dividing each eigenvector by the square root of the corresponding eigenvalue ., These whitened 100-dimensional inputs were used to train a network with 380 output neurons ., The results were robust to different manipulations of the inputs ., For example , we obtained qualitatively similar results even without dimensionality reduction or whitening , using smaller image patches ., The feed-forward filters were set to be Gabor filters with the same center in the visual field and the same spatial frequency ., The size of each Gabor filter was 25-by-25 pixels ., The full feed-forward matrix was a product of two matrices: A 380-by-625 matrix containing a Gabor filter in each row , which was multiplied from the right by a 625-by-100 matrix representing the reconstruction after the dimensionality reduction ., Close to the critical point , accurate simulation of the network dynamics requires a long time due to the phenomenon of critical slowing down ., To explore the characteristics and dynamics of the network as it approached the critical point , we allowed simulations to run for very long periods ., Thus , simulations could take up to weeks to complete based on network size and the value of the learning rate ., When the evolving networks approached a critical point , the objective function tended to be very sensitive to changes in the pattern of interactions ., In some cases , the objective function could even increase rather than decrease , implying that the learning rate was not small enough ., To overcome this problem , we calculated the expected value of the objective function before actually updating the interactions ., When an upcoming increase was identified , the learning rate was reduced by a factor of one-half and the process was repeated again ., The architecture of the network model is presented in Fig 1B ., Each input sample is a point on the plane , with an angle , θ0 , representing the orientation of a visual stimulus and amplitude ( its distance from the origin ) , r , representing the timulus contrast ( Fig 1B ) ., Each point can be represented as ( x1 , x2 ) = r ( cosθ0 , cosθ0 ) ., For clarity , we consider periodicity of 360° rather than 180° , which is the relevant symmetry when considering orientations ., The angles θ0 are distributed uniformly between 0 and 2π ., The amplitudes r are distributed according to a Gaussian distribution with a positive mean 〈r〉 , representing the mean contrast ., By varying the mean value of r we study the effect of stimulus statistics on the optimal network connections ., The network represents this two-dimensional input by M sigmoidal neurons ( M≫1 ) interconnected with recurrent interactions ( Kij , i , j = 1 , … , M ) ., The feedforward connections ( rows of ) are chosen to be unit vectors , uniformly distributed over all possible directions , i . e . ( Wi1 , Wi2 ) = r ( cosϕi , cosϕi ) where ϕi = 2πi/M , i = 1 , … , M ., Thus , the input to the i’th neuron has a cosine tuning function peaked at ϕi and the network has a ring architecture ( Fig 1B ) ., The feedforward connections are fixed throughout the learning ., Our goal is to evaluate the matrix of recurrent connections K that maximizes the mutual information between the steady state responses of the output neurons and the stimulus ., For a given input and connection matrix , the steady-state responses are given by, si=g ( ΣjKijsj+x1cosθi+x2sin\xa0θi ), ( 5 ), where g is the logistic function ( see Methods ) ., The sensitivity matrix , χ , is an M×2 matrix given by:, χi1=∂si∂x1=gi′⋅ΣlKilχl1+cosθi, ( 6 ), χi1=∂si∂x2=gi′⋅ΣlKilχl2+sinθi, ( 7 ), Where gi=g ( ΣjKijsj+x1cosθi+x2sin\xa0θi ) is the derivative function of the neuronal transfer function and we have used the expression for si given in Eq ( 5 ) ., To investigate analytically the optimal pattern of recurrent interactions when the typical input contrast is low , namely when 〈r〉→0 , we assume that the interaction Kij between the i’th and j’th neurons is an even function of the distance between the neurons on the ring ,, Kij=K ( θi−θj ), ( 8 ), When 〈r〉 approaches zero , the total external input to each neuron approaches zero ., We denote the value of g’ at zero input by γ0 = g’ ( 0 ) ., In the case of the logistic function , γ0 = 1/4 ., Since the number of output neurons , M , is large , we can take the continuum limit and transform the summations over angles to integrals ., For instance , the equation for χi1 can be written as, χ1 ( θ ) =γ0M2π∫−ππdθ′K ( θ−θ′ ) χ1 ( θ′ ) +cosθ, ( 9 ), and similarly for χi2 ., We define the Fourier series of K and χ1, K ( θ ) =1MΣn=0∞kncos ( nθ ), ( 10 ), χ1 ( θ ) =Σn=0∞ancos ( nθ ) +bnsin ( nθ ) , ( 11 ), Fourier transforming Eq ( 9 ) yields an=γ0δn1/ ( 1−12γ0k1 ) and bn = 0 , where k1 is the first cosine harmonic of the interaction profile , Eq ( 10 ) ., Thus ,, χi1=γ0cosθi1−12γ0k1, ( 12 ), and similarly, χi2=γ0sinθi1−12γ0k1, ( 13 ), The 2 X 2 matrix χTχ is a diagonal matrix with elements, ( χTχ ) 11= ( χTχ ) 22=Mγ02 ( 1−12γ0k1 ) 2, ( 14 ), Substituting these expressions in Eq ( 3 ) , yields, ε=log ( 2Mγ02 ) +2log ( 1−12γ0k1 ), ( 15 ), Eq ( 15 ) implies that as k1 approaches the critical value k1c= 2/γ0 the objective function diverges to −∞ ., This means that the optimal pattern of recurrent interactions has the form, K ( θi−θj ) =2Mγ0cos ( θi−θj ), ( 16 ), The divergence of the objective function , that is of the sensitivity ( or susceptibility ) at k1c reflects the fact that at this point the network undergoes a phase-transition into a state of spontaneous symmetry breaking 9 ., Formally , this can be illustrated by adding a uniform random component to the input that each neuron receives and examining the network response ., As shown in 9 , the network response is very different below and above the transition point ., For k1c<2/γ0 , the network settles into a homogeneous state with si = g ( 0 ) ., However , for k1c>2/γ0 , the network dynamics evolve into an inhomogeneous solution with a typical \\hill\\ shape 9 , which is determined by the recurrent connections and can be interpreted as a hallucination of an oriented stimulus ., Neurons , which are slightly more active due to the random noise , enhance the activity of neurons with similar preferred orientations , which in turn enhance the activity of the initial neurons through feedback ., The winning neurons inhibit neurons with more distant preferred orientations , thus creating a hill-shaped profile ., The location of the peak of this hill is arbitrary and depends on the specific realization of the noise in the input pattern and on the initial conditions of the neuronal activities ., This dramatic change in the network behavior implies that near k1c the network is extremely sensitive to small changes in the input ., This enhanced sensitivity increases the mutual information between the network response and the stimulus ., In the limit of 〈r〉→0 the objective function depends solely on the first harmonics of the interaction profile , leaving open the question of whether the higher order corrections in r predict large values of the higher harmonics of the interaction profile ., Furthermore , in the analytic derivation we have assumed translational invariance of K , which raises the question of whether there are better solutions which break this symmetry of K . To address these questions , we simulated the gradient based learning algorithm for the evolution of the interaction matrix ( Methods; 36 ) , with no restrictions on the form of the matrix ., The network consisted of 2 input neurons and 141 output neurons ., The nonlinear squashing function was the logistic function ., The feedforward connections to each output neuron were unit vectors uniformly distributed between 0° and 360° , and were fixed throughout the learning ., The initial recurrent interaction matrix was set to zero ., The angle of each input was drawn from a uniform distribution , while the magnitude was drawn from a Gaussian distribution around a characteristic radius r with a standard deviation of 0 . 1 times the mean ., Fig 2 shows the results from a simulation with 〈r〉 = 0 . 1 , namely when the inputs are relatively weak ., As can be seen , the interaction pattern ( Fig 2A ) is translation invariant; i . e . , each neuron has the same pattern of pre and postsynaptic interactions ., It is important to note that we do not impose any symmetry on the connections ., The resulting translation invariance is a natural result of the statistical symmetry of the inputs to the network ., Fig 2B shows one row of the interaction matrix ( representing the presynaptic connections into a single output neuron ) ., For clarity , the values are multiplied by the number of neurons , M . This result is highly congruent with the analytical derivation presented above , Eq ( 16 ) , that predicts a pure cosine profile with an amplitude of 8 for the logistic function ., Fig 2C shows the response of the network as a function of the preferred orientation ( PO ) of the neurons ( solid line ) to a vertical input at the typical contrast ( r = 0 . 1 ) ., The amplification in comparison to the network response without recurrent interactions ( dashed line ) is clearly seen ., Responses to different contrasts are shown in Fig 2D ., We next investigated a more complex network model of a visual hypercolumn ( Fig 1C ) ., In this setting , gray-level image patches from natural scenery ( see Methods ) were used as inputs to train the network 40 ., The network consisted in this case of 100 input neurons and 380 output neurons ., To study the pattern of recurrent interactions systematically , we manually set the feed-forward filters to be Gabor filters with the same center in the visual field and the same spatial frequency , spanning all orientations ., It is worth noting that this overcomplete network can also be used to learn the feed-forward connections themselves 36 , and indeed , as we established thorough numerical simulations , when trained using natural scenes , the feed-forward filters turn out to be Gabor-like filters ., This result is related to the fact that the algorithm for the feed-forward connections is a simple generalization of the infomax ICA algorithm 32 from complete to overcomplete representations ., Training the infomax ICA algorithm using natural scenes is known to result in Gabor-like filters 33 ., Fig 4A depicts the full matrix of recurrent connections ., As can be seen , the matrix is symmetric and the interaction between two neurons depends only on the distance between their preferred orientations ., This finding is in line with the behavior of the simple toy model ., Again , it is important to note that the interaction matrix was not constrained to be symmetric ., Rather , this is a natural outcome of the learning process , reflecting the symmetry in the pattern of feedforward interactions ., Fig 4B plots the interaction strength as a function of the distance between the preferred orientations of the pre- and post-synaptic neurons ., The emerging profile has a Mexican hat shape , with short-range excitation , longer-range inhibition and an oscillatory decay as the distance in preferred orientation increases ., To characterize the network behavior after training it with natural images we examined its response to simple oriented stimuli ., Fig 4C depicts the steady-state profile of activity in response to a vertically oriented Gabor stimulus ( solid line ) ., The spatial frequency of the Gabor stimulus and the width of the Gaussian envelope were identical to those of the Gabor filters in the feedforward connections and the contrast was set to the mean contrast of the training stimuli ., For comparison , the dashed line shows the response of the network without recurrent interactions ., Clearly , the evolved recurrent interactions amplify and sharpen the response compared to the response without recurrent interactions ., Fig 4D shows the network response to the same vertical stimulus for various contrast levels ., Notably , the width of the profile is approximately independent of the contrast , and the effect of changing the contrast is mainly multiplicative ., Fig 4E–4G show the dependence of various measures for the network behavior ( see Methods ) on the scaling factor ., Fig 4E shows that even small changes to the scale factor can significantly increase the objective function , resulting in poor information representation ., Decreasing the scale factor reduces the amplification provided by the recurrent interactions and consequently reduces the sensitivity of the network to external inputs ., Conversely , increasing the scale factor to values above 1 causes the recurrent interactions to become too dominant , and pushes the network into a pattern formation regime ., In this regime , the network is again less sensitive to external inputs , but this time it is due to the attractor dynamics that govern its behavior ., Fig 4F shows the convergence time of the network dynamics ., At the optimal point , the convergence time starts to increase to very high values , reflecting critical-slowing down at the transition into attractor-dominated dynamics ., The magnitude of the population vector also rises sharply near the optimal point ( Fig 4G ) ., Overall , the behavior of the convergence time and the population vector shows that indeed close to the optimal scaling factor from the learning process , the network experiences a phase transition ., The behavior of these metrics also resembles their behavior in the low contrast case in the toy model ( Fig 2F–2G ) ., The study focused on a simplified model of visual hypercolumn , a local processing unit in the visual cortex ., The feedforward interactions from the input layer to the output layer were manually set such that each neuron in the output layer had a certain preferred orientation ., The recurrent interactions among these neurons evolved according to learning rules that maximize the mutual information between the external input to the network and the networks steady-state output ., When the inputs to the network during learning were natural images , the evolved profile of interactions had a Mexican-hat shape ., The idea that neurons with similar preferred orientations should effectively excite each other and that neurons with distant preferred orientations should effectively inhibit each other has been suggested in the past based on empirical findings , e . g . 9 , 41 , 42 , but here it was derived using a first-principle computational approach ., This pattern of interactions helps in amplifying the external inputs and in achieving a relatively constant width for the orientation tuning curves , which is consistent with experimental findings on primary visual cortical neurons 43 , 44 ., A learning algorithm for information maximization in recurrent neural networks was also derived in 34 ., The major difference from the current work is that here the information is maximized between the external input and the steady-state output , whereas in 34 the input and output refer to the patterns of activity in the recurrent network at two consecutive time steps ., The approach in 34 is aimed at maximizing information retention in the recurrent network , whereas here the focus is on sensory processing and on the representation of the external input ., In addition , the neurons in 34 are stochastic binary neurons , whereas the neurons here are deterministic and have a smooth nonlinearity ., The network model in 34 was also trained using natural images as external inputs , leading to Gabor-like feed-forward connections , consistent with the findings in 33 ., However , the authors do not discuss the structure of the connections among the output neurons , so this important aspect cannot be compared with the present work , which focused on recurrent connectivity ., The present model is clearly overly simplified in many aspects as a model of the primary visual cortex ., For example , the gradient-based learning rules employed here are likely to be very different from the plasticity mechanisms in the biological system , but the assumption is that they reflect the long-term evolution of the relevant neural system and converge to a similar functional behavior ., Despite its simplicity , the model provides a concrete setting for examining the role of recurrent interactions in the context of sensory processing ., This leads to general insights that go beyond the context of early visual processing , as we discuss below ., The dynamics of recurrent networks , like the one studied here , can allow the network to hold persistent activity even when the external drive is weak or absent ., The network is then said to display attractor dynamics ., In the context of memory systems , attractors are used to model associative memory 45 , 46 ., Different attractors correspond to different memory states , and the activity patterns that form the basin of attraction of each attractor correspond to various associations of this memory ., In the context of early sensory networks , however , the persistent activity at an attractor may correspond to a hallucination ., In addition , the flow from different initial patterns to the attractor implies loss of information and insensitivity to changes in the external inputs , and thus may be undesired in the context of sensory processing ., An important result of this study is that the evolved networks naturally tend to operate near a critical point , which can be thought of as the border between normal amplification of inputs and hallucinations ., In 9 , a model of a visual hypercolumn , which is similar to our toy model , was studied analytically ., There , the pattern of interactions was assumed to have a cosine profile and it was shown that when the amplitude of the cosine crosses a critical value , the network transitions into an attractor regime ., In this regime , the network dynamics evolve into an inhomogeneous solution with a typical hill shape , which represents a hallucination of an oriented stimulus ., Here , the learning algorithm leads the network to operate close to that critical point ., Scaling up the resulting pattern of synaptic interactions by a small factor pushes the network into the undesired regime of attractors , namely into hallucinations 47 , 48 ., This tendency to operate near a critical point can be explained intuitively ., The task of the network is to maximize the mutual information between input and output , which amounts to maximizing its sensitivity to changes in the external inputs ., The network uses the recurrent interactions to amplify the external inputs , but too strong amplification may generate hallucinations ., Thus , the learning process should settle at an optimal point , which reflects a compromise between these two factors ., An interesting insight comes from comparing the network to physical systems that may experience phase-transitions in their behavior ., A universal property of these systems is that their sensitivity to external influences , or in physical terminology their susceptibility , is maximized at the transition point 49 ., Our adaptive sensory recurrent networks evolve to operate near a critical point in order to achieve maximal susceptibility and represent information optimally ., It is important to note that neural systems respond to a wide range of inputs and that the target of the learning is to find the pattern of interactions that is optimal on average ., Under certain conditions , the recurrent interactions may not contribute much to the representation ., However , in many cases , especially if the typical inputs have a narrow distribution or tend to be weak , the optimal pattern of recurrent interactions is expected to be near critical ., The dominance of low contrasts in natural images is therefore an important factor in driving the pattern of recurrent interactions to be near critical ., There are several important distinctions to be made when comparing previous research 14 , 15 , 24 , 27 , 50 , 51 on critical brain dynamics with the present study ., First , the present work addresses mainly the issues of long-term plasticity and the effect of input statistics , whereas previous modeling works consider mostly networks with random connectivity , which do not adapt to input statistics ., Here we demonstrated that near-criticality emerges as a result of directly optimizing a well-defined measure for network performance using a concrete learning algorithm ., In addition , an important role is played by the input statistics , and depending on these statistics the network may or may not approach criticality ., Moreover , the resulting connectivity matrices are not random and the specific pattern that emerges is crucial for the network performance ., We note that in 34 the network can adapt to the statistics of external inputs , but there criticality was demonstrated only when the network evolved without external input ., Other studies , such as 52 , model plasticity in recurrent neuronal networks , but not in an ecological sensory context ., Second , here the critical point relates to the transition from normal amplification of external inp | Introduction, Methods, Results, Discussion | Recurrent connections play an important role in cortical function , yet their exact contribution to the network computation remains unknown ., The principles guiding the long-term evolution of these connections are poorly understood as well ., Therefore , gaining insight into their computational role and into the mechanism shaping their pattern would be of great importance ., To that end , we studied the learning dynamics and emergent recurrent connectivity in a sensory network model based on a first-principle information theoretic approach ., As a test case , we applied this framework to a model of a hypercolumn in the visual cortex and found that the evolved connections between orientation columns have a Mexican hat profile , consistent with empirical data and previous modeling work ., Furthermore , we found that optimal information representation is achieved when the network operates near a critical point in its dynamics ., Neuronal networks working near such a phase transition are most sensitive to their inputs and are thus optimal in terms of information representation ., Nevertheless , a mild change in the pattern of interactions may cause such networks to undergo a transition into a different regime of behavior in which the network activity is dominated by its internal recurrent dynamics and does not reflect the objective input ., We discuss several mechanisms by which the pattern of interactions can be driven into this supercritical regime and relate them to various neurological and neuropsychiatric phenomena . | The recurrent interactions among cortical neurons shape the representation of incoming information but the principles governing their evolution are yet unclear ., We investigate the computational role of recurrent connections in the context of sensory processing ., Specifically , we study a neuronal network model in which the recurrent connections evolve to optimize the information representation of the network ., Interestingly , these networks tend to operate near a critical point in their dynamics , namely close to a phase of hallucinations , in which non-trivial spontaneous patterns of activity evolve even without structured input ., We provide insights into this behavior by applying the framework to a network of orientation selective neurons , modeling a processing unit in the primary visual cortex ., Various scenarios , such as attenuation of the external inputs or increased plasticity , can lead such networks to cross the border into the supercritical phase , which may manifest as neurological and neuropsychiatric phenomena . | learning, medicine and health sciences, neural networks, brain, social sciences, neuroscience, learning and memory, cognitive psychology, network analysis, hallucinations, vision, neuronal tuning, computer and information sciences, animal cells, visual cortex, psychology, cell biology, anatomy, neurons, biology and life sciences, cellular types, sensory perception, cognitive science | null |
Subsets and Splits